NSGA III-based multi-unmanned aerial vehicle power inspection scheduling method

Abstract

The invention relates to a multi-unmanned aerial vehicle electric power inspection scheduling method based on NSGA III, and belongs to the field of unmanned aerial vehicle electric power inspection scheduling. According to the method, firstly, an electric power inspection task is modeled as a multi-objective optimization problem, two optimization objectives of total risk and maximum risk are considered, and multiple risk factors such as unmanned aerial vehicle operation risk, inspection point approaching facility risk and unmanned aerial vehicle obstacle leap facility risk are introduced. And performing optimization scheduling on the inspection task through an NSGA-III algorithm to realize efficient and collaborative electric power inspection. Compared with a rule-based algorithm and the algorithm provided by the invention, the method has remarkable advantages in the aspects of reducing the total risk and the maximum risk. Especially in an unmanned aerial vehicle failure continuing scene, the task of the failed unmanned aerial vehicle can be quickly and effectively replaced by the other unmanned aerial vehicle, and the continuity and integrity of routing inspection are ensured. The method provides powerful technical support for efficient and safe execution of electric power inspection tasks, and has a wide application prospect and an important practical value.

Description

Technical Field

[0001] The present invention belongs to the field of UAV power inspection and dispatching, and in particular relates to a multi-UAV power inspection and dispatching method based on NSGAIII. Background Art

[0002] With the rapid development of the power industry and the continuous expansion of the power grid, in order to ensure the safe and stable operation of overhead lines, power line inspection is a very important means to protect the normal operation of the lines. However, the traditional manual power line inspection method is time-consuming, costly, and has potential risks due to insufficient safety measures. In contrast, the use of unmanned aerial vehicles (UAVs) equipped with cameras and lidar sensors to perform power line inspection tasks can demonstrate efficient, safe and flexible operation capabilities in various environments, bad weather and complex terrain. Due to the limited endurance of a single drone, it is impossible to efficiently complete a large range of power inspection tasks. The collaborative operation of multiple drones can often complete power inspection tasks in a large range of scenarios more efficiently. Especially in the case of collaborative operation of multiple drones, work efficiency can be significantly improved, and reasonable planning of the inspection order of each drone plays a key role in optimizing overall efficiency and reducing costs. Summary of the invention

[0003] The purpose of the present invention is to provide a multi-UAV power inspection scheduling method based on NSGAIII for the optimization scheduling problem of multi-UAV collaborative operation in power inspection tasks. This method uses UAVs to perform power inspection tasks in order to solve the problems of time-consuming, high cost and insufficient safety measures in traditional power inspection methods. The multi-UAV power inspection is optimized and scheduled by the NSGA-III algorithm, especially when a UAV fails in the UAV operation sequence, the remaining UAVs can be reasonably scheduled to complete the inspection operation, optimize the total risk and the maximum risk of a single risk factor, and meet the maximum flight time constraint of the UAV, so as to complete the power inspection task efficiently and safely.

[0004] To achieve the above purpose, the technical solution of the present invention is: a multi-UAV power inspection and dispatching method based on NSGAIII, comprising:

[0005] Step 1: Model the power inspection task and identify multiple risk factors, including the risk of drone operation, the risk of approaching inspection point facilities, and the risk of drone flying over obstacles.

[0006] Step 2: Set optimization goals, including minimizing the total risk and minimizing the maximum risk of a single risk factor;

[0007] Step 3: Use the NSGA-III algorithm to optimize the scheduling of multi-UAV power inspection tasks, where the NSGA-III algorithm includes population initialization, selection operation, crossover and mutation operation, population merging and environment selection steps. The environment selection step involves non-dominated sorting and reference point selection mechanism;

[0008] Step 4: Based on the optimization results, generate inspection paths and inspection sequences for multiple drones to ensure that each inspection point is inspected at least once and the flight time of the drone is within the preset maximum flight time limit.

[0009] The present invention also provides a multi-UAV power inspection and dispatching system based on NSGAIII, comprising:

[0010] Modeling module, used to model power inspection tasks and determine multiple risk factors;

[0011] An optimization target setting module is used to set minimizing total risk and minimizing maximum risk as optimization targets;

[0012] The optimization scheduling module uses the NSGA-III algorithm to optimize the scheduling of multiple UAV power inspection tasks and generate inspection paths and inspection sequences for multiple UAVs;

[0013] The constraint checking module is used to ensure that each inspection point is inspected at least once and the flight time of the drone is within the preset maximum flight time limit.

[0014] Compared with the prior art, the present invention has the following beneficial effects: The present invention mainly solves the problem of continuous multi-UAV inspection after a UAV fails during the execution of a power inspection task. By applying the NSGA-III algorithm, the present invention can re-plan the remaining UAVs after a certain UAV fails to ensure that they can continue to complete the established inspection tasks. This solution not only significantly reduces the total risk in the inspection process and the maximum risk of a single risk factor, but also maintains the stability and controllability of the risk, thereby effectively improving the efficiency and reliability of power inspections. Compared with traditional rule-based algorithms, the present invention shows stronger adaptability and robustness when dealing with emergencies such as UAV failures, providing a strong guarantee for the smooth completion of UAV power inspection tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flow chart of the present invention;

[0016] Figure 2 This is a schematic diagram of the NSGA-III algorithm module of the present invention;

[0017] Figure 3 It is a scatter plot comparison of the present invention and the traditional rule-based algorithm when optimizing multiple objectives;

[0018] Figure 4 This is a comparison chart of the effects of the present invention and the traditional rule-based algorithm on total risk;

[0019] Figure 5 This is a comparison chart of the effects of the present invention and the traditional rule-based algorithm when the risk is the maximum for a single risk factor;

[0020] Figure 6 A comparison diagram of the total risk effect of the present invention and the traditional rule-based algorithm in the case of failure continuation;

[0021] Figure 7 The figure is a comparison diagram of the total risk effect of the present invention and the traditional rule-based algorithm in the case of failure continuation. DETAILED DESCRIPTION

[0022] The technical solution of the present invention is described in detail below in conjunction with the accompanying drawings.

[0023] Embodiment 1:

[0024] The present invention is achieved through the following technical solutions: Figure 1 As shown, a multi-UAV power inspection and dispatching method based on NSGA-III includes the following steps:

[0025] Step 1: Model the power inspection task and determine multiple risk factors, including the risk of drone operation, the risk of approaching inspection point facilities, and the risk of drone flying over obstacles.

[0026] First, R1(x i ), that is, the operational risk of the drone. This risk factor is closely related to the actual operational risk of the drone, mainly considering its flight time and speed. In the setting, for every additional second of flight time, the risk it faces increases by 0.0001, that is, the instantaneous coefficient of flight risk is 0.0001 per second. Assume that the drone flies at a constant speed of 5 meters per second and stays at each preset inspection point for 20 seconds to perform necessary inspection tasks. Based on this, for any drone numbered i, its operational risk R1(x i ) will be calculated strictly according to its actual flight time and stay time in the inspection sequence xi.

[0027] R1(x i )=0.0001t,t≥0

[0028] The second is R2(x i), that is, the risk of approaching the inspection point facilities. In this problem, the present invention considers the risk of the inspection point itself, which is mainly determined by the environmental conditions around it, mainly including the risk of approaching power lines and towers. For the risk related to the tower, its quantitative expression is:

[0029]

[0030] Where d is the distance between the drone and the nearest tower, in meters. Similarly, the risk associated with power lines is quantified by the following expression:

[0031]

[0032] Where d is the distance between the drone and the nearest wire, in meters. After integration, we can get:

[0033] R2(x i )=R2a(x i )+R2b(x i ),i=1,2,...,n

[0034] Finally, R3(x i ), that is, the risk of the drone flying over obstacles. During the inspection mission, the drone may need to fly over obstacles such as power lines or poles, which themselves contain risks. In this problem, the risk of flying over power lines is set to 0.005, while the risk of flying over poles is higher, reaching 0.1. The quantitative expression is as follows:

[0035]

[0036] Step 2, setting optimization goals, including minimizing the total risk and minimizing the maximum risk of a single risk factor;

[0037] Now consider transforming the complex task of multi-UAV power inspection into a problem involving multiple objectives and subject to multiple constraints, namely a multi-objective optimization problem. The standard mathematical model of a multi-objective optimization problem can generally be expressed as:

[0038] miny=F(x)=(f1(x),f2(x),...,f k (x)

[0039]

[0040] where g i (x) and h j (x) are I inequality constraints and J equality constraints, x=(x1,x2,...,x n ) is an n-dimensional decision vector, f k(x) is the k-dimensional design target.

[0041] Assume that in the UAV inspection power line mission scenario, there are a total of n inspection points that need to be monitored, and 3 UAVs are equipped to perform the inspection mission. For each UAV, we use the variable x i To represent its inspection sequence, where i represents the number of the drone (i = 1, 2, 3). These inspection sequences x i The maximum length of can reach n, which means that each drone can theoretically inspect all n inspection points in sequence. i The value range of the elements in is limited to between 1 and n, and each specific value k (k∈{1,2,...,n}) represents the serial number of a specific inspection point inspected by the drone.

[0042] The present invention has two optimization goals, the total risk F1(x) and the maximum risk F2(x) of a single risk factor. The total risk aims to minimize the risk faced by the drone during the entire inspection process, while the maximum risk of a single risk factor focuses on the maximum risk value that the drone may encounter during the inspection process. In some cases, even if the total risk is low, serious consequences may occur if there is a high risk point. Its quantitative expression is as follows:

[0043]

[0044]

[0045] Step 3: Use the NSGA-III algorithm to optimize the scheduling of inspection tasks. The NSGA-III algorithm includes population initialization, selection operation, crossover and mutation operation, population merging, and environment selection steps. The environment selection step involves non-dominated sorting and reference point selection mechanisms.

[0046] Step 4: Generate inspection paths and inspection sequences for multiple drones based on the optimization results to ensure that each inspection point is inspected at least once and that the flight time of the drone is within the preset maximum flight time limit. Set the maximum flight time of the drone to 500 seconds.

[0047] Figure 2 This is the flow chart of the NSGA-III algorithm. The NSGA-III algorithm process is as follows:

[0048] 1) Initialize the population: Randomly generate an initial population containing multiple individual solutions. Each individual is usually represented by a set of decision variables.

[0049] 2) Non-dominated sorting: Perform non-dominated sorting on the individuals in the initial population and divide them into different levels. The first level contains non-dominated solutions on the Pareto front, the second level contains solutions dominated by the first level, and so on.

[0050] 3) Calculate the crowding degree: In order to measure the density of individuals in the frontier, the crowding degree of each solution is calculated to maintain diversity during the selection process.

[0051] 4) Selection operation: Select a portion of solutions to create the next generation population. During the selection process, all solutions in the first few non-dominated levels will be selected to ensure that high-quality solutions are retained. For solutions of the same level, a portion will be selected based on their crowding to maintain diversity. The selected solutions will enter the next generation population.

[0052] 5) Crossover and mutation operations: Use the crossover operation to generate one or more child solutions from two parent solutions. The crossover operation usually uses the simulated binary crossover (SBX) method. Use the mutation operation to make small random changes to the individual decision variables. The mutation operation usually uses the polynomial mutation method.

[0053] 6) Merge population: merge the newly generated solution with the previous generation population to form a new 2N population, where N is the initial population size.

[0054] 7) Environment selection: If the population size exceeds the required size, a selection strategy is used to select the most suitable solution. It is necessary to use non-dominated sorting and reference point selection mechanisms to maintain the diversity of the population by introducing widely distributed reference points. These reference points are used to measure the superiority of the solution in the target space and help select solutions that can be evenly distributed throughout the target space.

[0055] 8) Check the termination condition: Determine the set termination condition, such as the maximum number of iterations or other convergence criteria. If the termination condition is not met, return to step 2 and continue the next generation of evolution. If the termination condition is met, output the result.

[0056] 9) Output results: Output the set of high-quality solutions on the Pareto frontier found as the output of the algorithm.

[0057] Figure 3This is a scatter plot comparison of the present invention and the traditional rule-based algorithm when optimizing multiple objectives. At present, power companies mainly use rule-based algorithms when conducting power inspections. This algorithm is a computational method that solves problems based on pre-set rules and conditions. These rules and conditions may be expressed as rigorous logical relationships or as a series of detailed steps and operating procedures. Such algorithms operate in a deterministic manner and can be clearly explained and implemented. In the actual application of power inspections, power companies usually use this algorithm to assign inspection tasks to drones based on the principle of proximity.

[0058] As can be seen from the picture, among the many scattered points, excluding some noise interference items, the algorithm we proposed is significantly better than the rule-based algorithm in both maximum risk and total risk. Compared with the rule-based algorithm used in traditional power inspection, our algorithm can effectively maintain the trend of maximum risk not increasing while reducing the total risk. And when some scattered points of the two methods intersect, our algorithm can achieve a maximum risk lower than the rule-based algorithm when the total risk is the same. When the maximum risk is basically the same, the algorithm we proposed can also achieve a total risk lower than the rule-based algorithm.

[0059] Depend on Figure 4 and Figure 5 The two figures show that the total risk and maximum risk of our algorithm under each scene sequence are generally lower than those of the rule-based algorithm, which shows that our proposed algorithm has better performance in reducing risk. The total risk and maximum risk of the two methods fluctuate with the change of the scene sequence number, which meets the complex applications in different inspection scenarios.

[0060] After setting a drone to fail during the inspection, we replanned the remaining two drones, continued the work on the task points that the failed drone had not completed, and used the NSGA-III algorithm to optimize it. Finally, we got Figure 6 and Figure 7 .Depend on Figure 6 and Figure 7 It can be seen that the algorithm we proposed is still significantly better than the rule-based algorithm. In terms of total risk, the volatility of the rule-based algorithm is more intense than that of the non-failure continuation, and the total risk is significantly increased. Although the total risk of the algorithm we proposed has also increased, the volatility is smaller. In terms of maximum risk, the rule-based algorithm has also increased significantly in some cases, while the algorithm we proposed has remained basically stable. It is worth noting that in the failure continuation scenario, the rule-based algorithm failed to meet the set 500-second constraint 128 times, which shows that its ability to cope with sudden failure conditions is also insufficient, while our algorithm meets the 500-second constraint.

[0061] The above are preferred embodiments of the present invention. Any changes made according to the technical solution of the present invention, as long as the resulting functions do not exceed the scope of the technical solution of the present invention, belong to the protection scope of the present invention.

Claims

1. A multi-UAV power inspection and dispatching method based on NSGAIII, characterized in that: include: Step 1: Model the power inspection task and identify multiple risk factors, including the risk of drone operation, the risk of approaching inspection point facilities, and the risk of drone flying over obstacles. Step 2: Set optimization goals, including minimizing the total risk and minimizing the maximum risk of a single risk factor; Step 3: Use the NSGA-III algorithm to optimize the scheduling of multi-UAV power inspection tasks, where the NSGA-III algorithm includes population initialization, selection operation, crossover and mutation operation, population merging and environment selection steps. The environment selection step involves non-dominated sorting and reference point selection mechanism; Step 4: Based on the optimization results, generate inspection paths and inspection sequences for multiple drones to ensure that each inspection point is inspected at least once and the flight time of the drone is within the preset maximum flight time limit.

2. According to a multi-UAV power inspection and dispatching method based on NSGAIII according to claim 1, it is characterized in that: The operational risk of the UAV is determined according to the flight time of the UAV and a preset instantaneous coefficient of flight risk.

3. According to a multi-UAV power inspection and dispatching method based on NSGAIII according to claim 1, it is characterized in that: The risk of approaching the inspection point facility is determined based on the distance between the drone and the nearest tower and a preset tower-related risk quantification expression.

4. The multi-UAV power inspection and dispatching method based on NSGAIII according to claim 1 is characterized in that: The risk of the drone flying over an obstacle facility is determined according to the type of obstacle that the drone flies over and a preset risk value of flying over the obstacle.

5. The multi-UAV power inspection and dispatching method based on NSGAIII according to claim 1 is characterized in that: It also includes solving the problem of failure continuation during the execution of drone inspection tasks, that is, when faced with the sudden situation of a drone failure, the remaining drones are re-planned, and the NSGA-III algorithm is used to optimize the re-planned inspection tasks to continue to complete the established inspection tasks.

6. A multi-UAV power inspection and dispatching method based on NSGAIII according to claim 1 or 2, characterized in that: The specific definition of the operational risks of the drone is as follows: R1(x i )=0.0001t,t≥0 Among them, R1(x i ) represents the operational risk of the drone, x i is the inspection sequence of the UAV numbered i, and t is the flight time of the UAV.

7. A multi-UAV power inspection and dispatching method based on NSGAIII according to claim 1 or 2, characterized in that: The risks of facilities close to inspection points are specifically defined as follows: R2(x i )=R2a(x i )+R2b(x i ),i=1,2,...,n Among them, R2(x i ) represents the risk of approaching the inspection point facilities, R2a(x i ) represents the risk of approaching the tower, R2b(x i ) represents the risk of approaching power lines, x i is the inspection sequence of the drone numbered i, d represents the distance between the drone and the nearest tower / nearest power line, and n is the total number of drones.

8. A multi-UAV power inspection and dispatching method based on NSGAIII according to claim 1 or 2, characterized in that: The risk of drones flying over obstacles is specifically defined as follows: Among them, R3(x i ) represents the risk of the drone flying over obstacles, x i is the inspection sequence of the drone numbered i, and over line / overtower means that the drone flies over the power line / tower.

9. The multi-UAV power inspection and dispatching method based on NSGAIII according to claim 1 is characterized in that: The NSGA-III algorithm is used to optimize two objectives, namely the total risk F1(x) and the maximum risk F2(x) of a single risk factor, which are defined as follows: Among them, R1(x i ) represents the operational risk of the drone, R2(x i ) represents the risk of approaching the inspection point facilities, R3(x i ) represents the risk of the drone flying over obstacles, x i is the inspection sequence of the drone numbered i, and n is the total number of drones.

10. A multi-UAV power inspection and dispatching system based on NSGAIII, characterized in that: include: Modeling module, used to model power inspection tasks and determine multiple risk factors; An optimization target setting module is used to set minimizing total risk and minimizing maximum risk as optimization targets; The optimization scheduling module uses the NSGA-III algorithm to optimize the scheduling of multiple UAV power inspection tasks and generate inspection paths and inspection sequences for multiple UAVs; The constraint checking module is used to ensure that each inspection point is inspected at least once and the flight time of the drone is within the preset maximum flight time limit.

Images