Unmanned aerial vehicle dynamic 3D trajectory design method suitable for disaster scene

Abstract

The invention relates to an unmanned aerial vehicle dynamic 3D trajectory design method suitable for a disaster scene, and belongs to the field of unmanned aerial vehicle wireless communication. The method comprises the following steps: performing priority division on clusters constructed by ground nodes according to a disaster area condition, calculating a cluster information age growth coefficient according to the priority division, and constructing each cluster information age model; calculating the maximum transmission radius of each cluster head according to the cluster head transmission power and the channel model of the cluster head and the unmanned aerial vehicle; within the maximum transmission radius, modeling the optimal acquisition point problem of the unmanned aerial vehicle as a Markov decision model, and establishing and solving an optimization problem by taking the weighted sum of the propulsion energy consumption of the unmanned aerial vehicle and the average information age of the network as a target to obtain the optimal acquisition point of the unmanned aerial vehicle in each cluster; and selecting the optimal acquisition point of the cluster with the minimum target function value, namely the next hovering acquisition point of the unmanned aerial vehicle. According to the method, the disaster degrees of different areas are distinguished, the data of the seriously affected areas are preferentially acquired within limited energy consumption, and the data acquisition efficiency is improved.

Description

Technical Field

[0001] The invention belongs to the field of unmanned aerial vehicle wireless communications and relates to a method for designing a dynamic 3D trajectory of an unmanned aerial vehicle suitable for disaster scenarios. Background Art

[0002] As a flexible, efficient and rapidly deployable aerial platform, drones have gradually become an ideal choice for data collection in disaster environments such as forest fires and nuclear leaks. Using drones as aerial relays for data collection can effectively ensure the stable operation of wireless networks in disaster areas, extend their service life, and be of great help in post-disaster rescue, disaster assessment, resource allocation, etc.

[0003] In the prior art, drones usually use a uniform collection method to collect data in disaster scenarios, which cannot distinguish the degree of disaster in different areas. This causes drones to waste precious energy on collecting data in areas with less severe disasters, and fail to obtain key information in severely affected areas in a timely manner, delaying rescue operations. Drone data collection path planning is often not optimized, causing drones to consume a lot of energy during flight, shortening their flight time and being unable to perform tasks in disaster areas for a long time. Due to the lack of targeted data collection, the base station center is unable to obtain key information in severely affected areas in a timely manner, making it difficult to make timely and effective response measures, resulting in the deterioration of the disaster and the expansion of the affected areas.

[0004] In view of the shortcomings of the prior art, the present invention proposes a method for designing a dynamic 3D trajectory of an unmanned aerial vehicle suitable for disaster scenarios. Summary of the invention

[0005] In view of this, the purpose of the present invention is to provide a method for designing a dynamic 3D trajectory of a UAV suitable for disaster scenarios. While keeping the energy consumption of the UAV as low as possible, data of disaster-stricken areas of different priorities are collected to minimize the average information age of the entire wireless network, so that the base station center can timely understand the real-time situation of the severely affected areas and take necessary response measures; thus, disasters such as forest fires and nuclear leaks can be contained in a timely manner, prevented from spreading, and strengthen the control of disasters.

[0006] In order to achieve the above object, the present invention provides the following technical solutions:

[0007] A method for designing a dynamic 3D trajectory of a UAV suitable for disaster scenarios includes the following steps:

[0008] The steps for designing the flight trajectory of a drone entering a disaster area from the current time slot to the next collection target are as follows:

[0009] S1. In the current time slot, the drone first collects the surrounding environment information, defines the information age growth coefficient of each cluster according to the disaster situation transmitted by the cluster and the average data generation cycle, and establishes the information age model;

[0010] S2, establish the UAV energy consumption model and channel model, and calculate the maximum transmission radius of each cluster head according to the channel model;

[0011] S3. Determine the hovering collection points of the UAV. Within the transmission radius of each cluster head, take the weighted sum of the UAV energy consumption and the average information age of the sensor nodes as the minimum, establish an optimization problem, and solve it to obtain the hovering collection points of each cluster head, and record the hovering collection points of each cluster and their corresponding objective function values.

[0012] S4. Compare the objective function values ​​of each cluster, and select the best collection point of the cluster when the objective function value is the smallest as the next collection position of the UAV, that is, the hovering collection point.

[0013] After the UAV completes all acquisition tasks, the trajectory formed by all hovering points is the three-dimensional flight trajectory of the UAV.

[0014] Furthermore, in S1, the priority is divided according to the disaster situation of the clusters and the information age growth coefficient of the clusters is calculated to establish an information age model, the specific contents of which are:

[0015] First, the location coordinates of each cluster member are obtained to cluster the sensor nodes; secondly, the K-means clustering algorithm is used to obtain the centroid of each cluster, and the sensor node with the shortest Euclidean distance to the centroid is used as the cluster head of the cluster. The flight trajectory of the drone is discretized into small time slots of length T. Assuming that the current time slot is n, at this time, the drone collects basic information about the nearby disaster-stricken area from the cluster head, including the M clusters detected within the transmission range and the severity of the disaster in the area where each cluster is located.

[0016] In order to reduce the loss of life and property in the disaster-stricken area, the disaster situation of each detected cluster is evaluated and the collection priority is determined. To characterize the degree of disaster, the priority of the i-th (i=1,2,...,M) cluster in time slot n is defined as The priority of cluster i The comprehensive disaster condition parameter F of the response cluster received by the UAV i (n) Decisions, including casualties Property damage temperature Smoke density represents the ratio of the number of casualties in the ith cluster at time slot n to the total number of casualties in the M clusters detected, represents the ratio of the property loss of the ith cluster in time slot n to the total property loss of the M detected clusters, represents the ratio of the average temperature detected by each node in the i-th cluster at time slot n to the maximum reference temperature, It represents the ratio of the average smoke concentration detected by each node in the i-th cluster at time slot n to the reference smoke concentration. The comprehensive disaster parameter F of cluster i i (n) is determined by the above four parameters and is expressed as:

[0017]

[0018] in and Represent the weights of casualties, property losses, temperature and smoke density parameters respectively, and satisfy

[0019] According to the disaster situation, the priority of clusters is divided into 5 levels. The priority of cluster i in time slot n is It is expressed as:

[0020]

[0021] In the above formula, the priority The larger it is, the more severe the disaster is in the area where the cluster is located, and the data of this cluster will have a greater chance of being collected by the drone first.

[0022] In order to reduce the average information age of wireless networks, in addition to the priority of clusters, the establishment of the information age model should also take into account the average data generation cycle of clusters. Assume that the i-th cluster contains K i cluster members, then the cluster member j (j=1,2,...,K i ) The data generation cycle in n time slots is expressed as Then the average data generation period of cluster i in time slot n is It can be calculated by the following formula:

[0023]

[0024] Further, define is the maximum average data generation period in M ​​clusters at time slot n, expressed as:

[0025]

[0026] Since the information age growth coefficient w of cluster i i(n) is related to the average data generation period of the clusters. To avoid the large difference in the growth coefficients between clusters, which may cause it to become the only decisive indicator for determining the trajectory of the drone, w i (n) is specifically expressed as:

[0027]

[0028] According to the cluster priority and information age growth coefficient, the information age model is established. In time slot n, the change of information age of data of member j of cluster i is expressed as follows:

[0029]

[0030] In the time slot when the cluster data is not collected by the drone, the information age of the cluster increases linearly, and in the time slot when the drone completes the data collection of the sensor node or the sensor node data is periodically updated, the information age of the sensor node will jump to the function value of the information age growth coefficient and priority of the cluster to which it belongs at the end of the time slot. It is worth noting that if node j is not scheduled to update the information age in a time slot, the information age at the end of the time slot and the beginning of the next time slot is continuous, otherwise it will experience a jump.

[0031] Furthermore, in S2, it is necessary to establish a UAV energy consumption model and a channel model, and calculate the maximum transmission radius of each cluster head according to the channel model. The specific contents are as follows:

[0032] Propulsion power of UAV in horizontal flight for:

[0033]

[0034] Among them, v H (n) is the horizontal flight speed of the UAV at time slot n, N u is the number of rotors of the UAV, c l is the rotor length of the UAV, ω u is the angular velocity, R u is the rotor disk radius, A u is the reference area of ​​the front end of the UAV, W u is the weight of the drone, ρ o and are air density and drag coefficient, λ u is a constant related to the physical quantity of the drone and the air density.

[0035] Assume that at time slot n, the drone is in a hovering state. At this time, the horizontal speed of the drone is v H (n)=0, then the hovering power of the drone is P H for:

[0036]

[0037] Affected by air resistance and gravity, the propulsion power required for the drone to ascend and descend in the vertical direction is different. If the vertical flight speed of the drone is v at time slot n, V (n), then the rising propulsion power and descent propulsion power The expressions are:

[0038]

[0039]

[0040] Since the transmission energy consumption of the UAV is too small compared to the propulsion energy consumption, it can be ignored. Therefore, in time slot n, the energy consumption model of the UAV is:

[0041]

[0042] Considering the impact of the signal propagation environment and the flight posture of the UAV on the air-ground channel in a disaster environment, a probabilistic path loss model is adopted for the line-of-sight link (LoS). In time slot n, the LoS probability between the UAV and cluster head i is It is expressed as:

[0043]

[0044] where a and b are constants that depend only on the environment, Indicates the coordinates of the drone (x(n), y(n), z(n)) in time slot n and the coordinates of cluster head i (x i ,y i ,z i ) is the Euclidean distance in the two-dimensional plane; h i (n)=|z(n)-z i | represents the vertical height difference between the UAV and cluster head i. Therefore, the probability of non-line-of-sight link NLoS can be calculated as:

[0045] At time slot n, the path losses of LoS and NLoS between the drone and cluster head i are:

[0046]

[0047]

[0048] Among them, f c is the carrier frequency, c is the propagation speed of light in vacuum, r i (n) is the three-dimensional Euclidean distance between the UAV and cluster head i, ηLoS and η NLoS denote the LoS and NLoS propagation losses, respectively. Therefore, at time slot n, the path loss L between the UAV and the ground cluster head i is i (n) is expressed as:

[0049]

[0050] In time slot n, cluster head i uses power When uploading data to the drone, the drone receiving power is expressed as In order to ensure that the drone can receive data reliably, the receiving power Should not be less than the threshold P min ,Right now It is expressed as:

[0051]

[0052] Substitute the link loss and corresponding probability under LoS and NLoS into the above formula for calculation, The transmission radius R of cluster head i can be obtained i (n) is:

[0053]

[0054] In the above formula, when the received power P of the drone r (n) = P min When the maximum transmission radius of cluster head i is obtained

[0055] Furthermore, in S3, in order to enable the drone to complete the collection task as quickly as possible with the minimum energy consumption, the weighted sum of network information age and drone energy consumption is minimized as the optimization goal. The specific contents are as follows:

[0056] According to the established information age model, the calculation expression of the average information age of the entire wireless network in n time slots is:

[0057]

[0058] in, is the total number of sensor nodes in the network.

[0059] Define δ1 and δ2 to represent the weights of the average information age of the wireless network and the energy consumption of the drone, respectively. Then the objective function is expressed as the weighted sum of the energy consumption of the drone and the network information age, that is: Assume that the cluster head acquisition strategy is G = {g1(n),g2(n),...,g M (n)},g i (n)∈{0,1} represents the indicator function of whether the UAV collects cluster head i data in time slot n. When gi When (n) = 0, it means that the UAV does not collect cluster head i. i When (n) = 1, the drone collects cluster head i. Considering the drone 3D trajectory S and cluster head collection strategy G, for any positive integer N, the following optimization problem is established:

[0060]

[0061]

[0062]

[0063]

[0064]

[0065]

[0066]

[0067]

[0068]

[0069] Among them, constraints C1, C2, and C3 respectively represent restrictions on the UAV's flight space; constraint C4 divides the priority of clusters; constraint C5 limits the elevation angle range between the UAV and the cluster head; constraints C6 and C7 respectively limit whether the UAV is scheduled and the scheduling rules; constraint C8 limits the maximum transmission radius of the cluster head.

[0070] Further, S3 is specifically:

[0071] S31, the drone determines whether the drone is within the maximum transmission radius based on the relative position of the current coordinates and the cluster head; if it is not within the maximum transmission radius, the drone coordinates are updated; if it is within the maximum transmission radius, the following steps are continued in this step;

[0072] S32, judging whether the current position is the hovering collection point of the cluster head according to the optimization objective function value, if not, continue to update the coordinates of the drone; if it is a hovering collection point, continue the following steps of this step;

[0073] S33, the drone sends a collection command, the cluster head sends data information to the drone, and updates the cluster information age;

[0074] Optionally, in S32, within the maximum transmission radius of each cluster head, a dual deep Q network algorithm (DDQN) is used to calculate the corresponding cumulative objective function values ​​of all possible actions, and the collection strategy and the optimal hovering point when the minimum value of the target cumulative function value is selected from all cluster heads according to the ε-greedy strategy. The optimal hovering point is the hovering collection point of the UAV in the next time slot.

[0075] The beneficial effects of the present invention are:

[0076] (1) By prioritizing the disaster-stricken areas and combining them with the information age model, drones can prioritize data collection in severely affected areas, obtain key information in a timely manner, provide strong support for rescue operations, and effectively improve data collection efficiency.

[0077] (2) By establishing the UAV energy consumption model and channel model and combining it with optimization problem solving and the DDQN algorithm, the UAV can optimize the flight path, reduce energy consumption, and extend the flight time, enabling it to perform tasks in the disaster area for a long time and ensure the continuity of data collection.

[0078] (3) Through optimization problem solving and DDQN algorithm, the UAV can complete the collection task with minimal energy consumption, while minimizing the average information age of the entire wireless network, improving the timeliness of information, and helping the base station center to understand the disaster situation in a timely manner and take necessary response measures, thereby curbing the spread of the disaster and strengthening the control of the disaster.

[0079] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0081] Figure 1 A schematic diagram of a model for drone-assisted data collection in disaster environments;

[0082] Figure 2 Schematic diagram of the information age of sensor node j in cluster i. DETAILED DESCRIPTION

[0083] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0084] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0085] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "rear", etc. indicate the orientation or position relationship, they are based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0086] See also Figure 1-2 , a dynamic 3D trajectory design method for UAVs suitable for disaster scenarios is as follows:

[0087] Step 1: Figure 1 As shown in the figure, when the drone enters the disaster area, it first obtains the position coordinates of each cluster member and clusters the sensor nodes; according to the disaster situation and urgency transmitted by the cluster, the clusters are divided into different priorities, and the information age growth coefficient of each cluster is defined according to the average data generation cycle, and the information age model is jointly established; in the three-dimensional flight space of the drone, there are M clusters within the detection range, and cluster i contains K inodes; the initial value of the information age of the node in the first time slot of the system starts to grow from zero. The K-means clustering algorithm is used to obtain the centroid of each cluster, and the sensor node with the shortest Euclidean distance to the centroid is used as the cluster head of the cluster. The flight trajectory of the drone is discretized into small time slots of length T. Assuming that the current time slot is n, at this time, the drone collects basic information about the nearby disaster-stricken area from the cluster head, including the M clusters detected within the transmission range and the severity of the disaster in the area where each cluster is located.

[0088] In order to reduce the loss of life and property in the disaster-stricken area, the disaster situation of each detected cluster is evaluated and the collection priority is determined. To characterize the degree of disaster, the priority of the i-th (i=1,2,...,M) cluster in time slot n is defined as The priority of cluster i The comprehensive disaster condition parameter F of the response cluster received by the UAV i (n) Decisions, including casualties Property damage temperature Smoke density represents the ratio of the number of casualties in the ith cluster at time slot n to the total number of casualties in the M clusters detected, represents the ratio of the property loss of the ith cluster in time slot n to the total property loss of the M detected clusters, represents the ratio of the average temperature detected by each node in the i-th cluster at time slot n to the maximum reference temperature, It represents the ratio of the average smoke concentration detected by each node in the i-th cluster at time slot n to the reference smoke concentration. The comprehensive disaster parameter F of cluster i i (n) is determined by the above four parameters and is expressed as:

[0089]

[0090] in and Represent the weights of casualties, property losses, temperature and smoke density parameters respectively, and satisfy

[0091] According to the disaster situation, the priority of clusters is divided into 5 levels. The priority of cluster i in time slot n is It is expressed as:

[0092]

[0093] In the above formula, the priority The larger it is, the more severe the disaster is in the area where the cluster is located, and the data of this cluster will have a greater chance of being collected by the drone first.

[0094] In order to reduce the average information age of wireless networks, in addition to the priority of clusters, the establishment of the information age model should also take into account the average data generation cycle of clusters. Assume that the i-th cluster contains K i cluster members, then the cluster member j (j=1,2,...,K i ) The data generation cycle in n time slots is expressed as Then the average data generation period of cluster i in time slot n is It can be calculated by the following formula:

[0095]

[0096] Further definition is the maximum average data generation period in M ​​clusters at time slot n, expressed as:

[0097]

[0098] Since the information age growth coefficient w of cluster i i (n) is related to the average data generation period of the clusters. To avoid the large difference in the growth coefficients between clusters, which may cause it to become the only decisive indicator for determining the trajectory of the drone, w i (n) is specifically expressed as:

[0099]

[0100] According to the priority of clusters and the information age growth coefficient, an information age model is established, such as Figure 2 As shown. In time slot n, the information age of the data of member j of cluster i is expressed as follows:

[0101]

[0102] In the time slot when the cluster data is not collected by the drone, the information age of the cluster increases linearly, and in the time slot when the drone completes the data collection of the sensor node or the sensor node data is periodically updated, the information age of the sensor node will jump to the function value of the information age growth coefficient and priority of the cluster to which it belongs at the end of the time slot. It is worth noting that if node j is not scheduled to update the information age in a time slot, the information age at the end of the time slot and the beginning of the next time slot is continuous, otherwise it will experience a jump.

[0103] Step 2: Establish the UAV energy consumption model and channel model, and calculate the maximum transmission radius of each cluster head based on the channel model. for:

[0104]

[0105] Among them, v H (n) is the horizontal flight speed of the UAV at time slot n, N u is the number of rotors of the UAV, c l is the rotor length of the UAV, ω u is the angular velocity, R u is the rotor disk radius, A u is the reference area of ​​the front end of the UAV, W u is the weight of the drone, ρ o and are air density and drag coefficient, λ u is a constant related to the physical quantity of the drone and the air density.

[0106] Assume that at time slot n, the drone is in a hovering state. At this time, the horizontal speed of the drone is v H (n)=0, then the hovering power of the drone is P H for:

[0107]

[0108] Affected by air resistance and gravity, the propulsion power required for the drone to ascend and descend in the vertical direction is different. If the vertical flight speed of the drone is v at time slot n, V (n), then the rising propulsion power and descent propulsion power The expressions are:

[0109]

[0110]

[0111] Since the transmission energy consumption of the UAV is too small compared to the propulsion energy consumption, it can be ignored. Therefore, in time slot n, the energy consumption model of the UAV is:

[0112]

[0113] Considering the impact of the signal propagation environment and the flight posture of the UAV on the air-ground channel in a disaster environment, a probabilistic path loss model is adopted for the line-of-sight link (LoS). In time slot n, the LoS probability between the UAV and cluster head i is It is expressed as:

[0114]

[0115] where a and b are constants that depend only on the environment, Indicates the coordinates of the drone (x(n), y(n), z(n)) in time slot n and the coordinates of cluster head i (x i ,y i ,z i ) is the Euclidean distance in the two-dimensional plane; h i (n)=|z(n)-z i | represents the vertical height difference between the UAV and cluster head i. Therefore, the probability of non-line-of-sight link NLoS can be calculated as:

[0116] At time slot n, the path losses of LoS and NLoS between the drone and cluster head i are:

[0117]

[0118] Among them, f c is the carrier frequency, c is the propagation speed of light in vacuum, r i (n) is the three-dimensional Euclidean distance between the UAV and cluster head i, η LoS and η NLoS denote the LoS and NLoS propagation losses, respectively. Therefore, at time slot n, the path loss L between the UAV and the ground cluster head i is i (n) is expressed as:

[0119]

[0120] In time slot n, cluster head i uses power When uploading data to the drone, the drone receiving power is expressed as In order to ensure that the drone can receive data reliably, the receiving power Should not be less than the threshold P min ,Right now It is expressed as:

[0121]

[0122] Substitute the link loss and corresponding probability under LoS and NLoS into the above formula for calculation, The transmission radius R of cluster head i can be obtained i (n) is:

[0123]

[0124] In the above formula, when the received power P of the drone r (n) = P min When , the maximum transmission radius of cluster head i can be obtained

[0125] Step 3: In order to enable the UAV to complete the collection task as quickly as possible with the minimum energy consumption, the weighted sum of network information age and UAV energy consumption is minimized as the optimization goal. According to the established information age model, the calculation expression of the average information age of the entire wireless network in n time slots is:

[0126]

[0127] in, is the total number of sensor nodes in the network.

[0128] Define δ1 and δ2 to represent the weights of the average information age of the wireless network and the energy consumption of the drone, respectively. Then the objective function is expressed as the weighted sum of the energy consumption of the drone and the network information age, that is: Assume that the cluster head acquisition strategy is G = {g1(n),g2(n),...,g M (n)},g i (n)∈{0,1} represents the indicator function of whether the UAV collects cluster head i data in time slot n. When g i When (n) = 0, it means that the UAV does not collect cluster head i. i When (n) = 1, the drone collects cluster head i. Considering the drone 3D trajectory S and cluster head collection strategy G, for any positive integer N, the following optimization problem is established:

[0129]

[0130] Among them, constraints C1, C2, and C3 respectively represent restrictions on the UAV's flight space; constraint C4 divides the priority of clusters; constraint C5 limits the elevation angle range between the UAV and the cluster head; constraints C6 and C7 respectively limit whether the UAV is scheduled and the scheduling rules; constraint C8 limits the maximum transmission radius of the cluster head.

[0131] Step 3 The specific content of the drone obtaining the hovering collection point according to the optimization target is:

[0132] In each time slot, the drone determines the target cluster by calculating the optimized objective function value, and executes the corresponding flight action to complete the coordinate update according to the target position decision action instruction. The three-dimensional coordinates of the drone in time slot n are expressed as dir(n) = (x(n), y(n), z(n)); after the drone makes a flight action strategy, the drone updates the coordinates according to the flight action instruction act(n) = (Δx(n), Δy(n), Δz(n)), which is specifically expressed as:

[0133] x(n+1)=x(n)+Δx(n)

[0134] y(n+1)=y(n)+Δy(n)

[0135] z(n+1)=z(n)+Δz(n)

[0136] dir(n+1)=(x(n+1),y(n+1),z(n+1))

[0137] Among them, Δx(n), Δy(n), and Δz(n) represent the differences in the horizontal plane and height that need to be transformed under the flight action command act(n). After the coordinate update is completed, the coordinate transformation record is stored. Step 3 can be divided into the following specific steps:

[0138] Step 3.1: The drone determines whether it is within the maximum transmission radius based on the relative position of the current coordinates and the cluster head; if it is not within the maximum transmission radius, the drone coordinates are updated; if it is within the maximum transmission radius, continue with the following steps in this step;

[0139] Step 3.2: Determine whether the current position is the hovering collection point of the cluster head according to the optimization objective function value. If not, continue to update the drone coordinates; if it is a hovering collection point, continue with the following steps in this step.

[0140] In particular, in step 3.2, within the maximum transmission radius of each cluster head, the dual deep Q network algorithm (DDQN) is used to calculate the corresponding cumulative objective function values ​​of all possible actions, and the collection strategy and the optimal hovering point when the minimum value of the objective cumulative function value is selected from all cluster heads according to the ε-greedy strategy. The optimal hovering point is the hovering collection point of the UAV in the next time slot.

[0141] Step 3.3: The drone sends a collection command, and the cluster head sends data information to the drone to update the cluster information age;

[0142] Step 4: Compare the objective function values ​​of each cluster, and select the best collection point of the cluster when the objective function value is the smallest as the next collection position of the drone, that is, the hovering collection point.

[0143] Step 5: After the drone completes all acquisition tasks, the trajectory formed by all hovering points is the three-dimensional flight trajectory of the drone.

[0144] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A method for designing a dynamic 3D trajectory of a UAV suitable for disaster scenarios, characterized by: The method comprises the following steps: S1: The drone enters the disaster-stricken area and designs a flight trajectory from the current time slot to the next hovering collection point. In the current time slot, the drone first collects information about the surrounding environment, defines the information age growth coefficient of each cluster based on the disaster situation transmitted by the cluster and the average data generation cycle, and establishes an information age model; S2: Establish the UAV energy consumption model and channel model, and calculate the maximum transmission radius of each cluster head based on the channel model; S3: Within the transmission radius of each cluster head, with the goal of minimizing the weighted sum of the drone energy consumption and the average information age of the sensor nodes, an optimization problem is established and solved to obtain the best collection point for each cluster, and the best collection point of each cluster and its corresponding objective function value are recorded; S4: Compare the objective function values ​​of each cluster, and select the best collection point of the cluster when the objective function value is the smallest as the next collection position of the drone, that is, the hovering collection point; After the UAV completes all collection tasks, the trajectory formed by all hovering collection points is the three-dimensional flight trajectory of the UAV.

2. The method for designing a dynamic 3D trajectory of an unmanned aerial vehicle suitable for disaster scenarios according to claim 1, characterized in that: In S1, the priority is divided according to the disaster situation of the clusters, the cluster information age growth coefficient is calculated, and the information age model is established. The specific contents are as follows: Each ground node broadcasts its own location information and performs clustering. The K-means clustering algorithm is used to obtain the centroid of each cluster, and the sensor node with the shortest Euclidean distance to the centroid is used as the cluster head of the cluster. The flight trajectory of the drone is discretized into small time slots of length T. Assuming that the current time slot is n, n = 1, 2, 3, ..., the drone collects basic information about the nearby disaster-stricken areas from the cluster head, including the M clusters detected within the transmission range and the degree of disaster in the area where each cluster is located. In order to reduce the loss of life and property in the disaster-stricken area, the disaster situation of each detected cluster is evaluated and the collection priority is determined; in order to characterize the degree of disaster, the priority of the i-th cluster in time slot n is defined as i=1,2,...,M; priority of cluster i The comprehensive disaster condition parameter F of the response cluster received by the UAV i (n) Decisions, including casualties Property damage temperature Smoke density represents the ratio of the number of casualties in the ith cluster at time slot n to the total number of casualties in the M clusters detected, represents the ratio of the property loss of the ith cluster in time slot n to the total property loss of the M detected clusters, represents the ratio of the average temperature detected by each node in the i-th cluster at time slot n to the maximum reference temperature, It represents the ratio of the average smoke concentration detected by each node in the i-th cluster at time slot n to the reference smoke concentration. The comprehensive disaster parameter F of cluster i i (n) is determined by the above four parameters and is expressed as: in and Represent the weights of casualties, property losses, temperature and smoke density parameters respectively, and satisfy According to the disaster situation, the priority of clusters is divided into 5 levels. The priority of cluster i in time slot n is It is expressed as: In the above formula, the priority The larger the value is, the more serious the disaster is in the area where the cluster is located, and the data of this cluster has a greater chance of being collected by drones first; In order to reduce the average information age of wireless networks, in addition to the priority of clusters, the establishment of the information age model should also take the average data generation period of clusters into consideration; Assume that the i-th cluster contains K i The data generation period of cluster member j in the i-th cluster group in n time slots is expressed as j=1,2,...,K i , then the average data generation period of cluster i in time slot n is It is calculated by the following formula: definition is the maximum average data generation period in M ​​clusters at time slot n, expressed as: Since the information age growth coefficient w of cluster i i (n) is related to the average data generation period of the cluster. In order to avoid the large difference in the growth coefficient between clusters, which may cause it to become the only decisive indicator for determining the trajectory of the drone, it is expressed in exponential form. i (n) is specifically expressed as: According to the priority of clusters and the information age growth coefficient, an information age model is established; in time slot n, the information age of member j in cluster i is It is expressed as: In the time slot when the cluster data is not collected by the drone, the information age of the cluster increases linearly. In the time slot when the drone completes the data collection of the sensor node or the sensor node data is periodically updated, at the end of the time slot, the information age of the sensor node will jump to the function value of the information age growth coefficient and priority of the cluster to which it belongs; It is worth noting that if node j is not scheduled to update its information age in a time slot, then the information age at the end of the time slot and the beginning of the next time slot is continuous, otherwise it will experience a jump.

3. The method for designing a dynamic 3D trajectory of a UAV suitable for disaster scenarios according to claim 2 is characterized in that: In S2, it is necessary to calculate the maximum transmission radius of each cluster head by establishing a system channel model, and the specific content is: Considering the impact of signal propagation environment and UAV flight posture on air-ground channel in disaster environment, a probabilistic path loss model is adopted for line-of-sight link LoS; at time slot n, the LoS probability between UAV and cluster head i is It is expressed as: where a and b are constants that depend only on the environment, Indicates the coordinates of the drone (x(n), y(n), z(n)) in time slot n and the coordinates of cluster head i (x i ,y i ,z i ) is the Euclidean distance in the two-dimensional plane; h i (n)=|z(n)-z i | represents the vertical height difference between the UAV and cluster head i; the probability of non-line-of-sight link NLoS is calculated as: At time slot n, the path losses of LoS and NLoS between the drone and cluster head i are: Among them, f c is the carrier frequency, c is the propagation speed of light in vacuum, r i (n) is the three-dimensional Euclidean distance between the UAV and cluster head i, η LoS and η NLoS denote the LoS and NLoS propagation losses respectively; therefore, at time slot n, the path loss L between the UAV and the ground cluster head i i (n) is expressed as: In time slot n, cluster head i uses power When uploading data to the drone, the drone receiving power is expressed as In order to ensure that the drone can receive data reliably, the receiving power Not less than the threshold P min ,Right now It is expressed as: Substitute the path loss and corresponding probability under LoS and NLoS into the above formula for calculation, Get the transmission radius R of cluster head i i (n) is: In the above formula, when the receiving power of the drone is When the maximum transmission radius of cluster head i is obtained 4. The method for designing a dynamic 3D trajectory of a UAV suitable for disaster scenarios according to claim 3 is characterized in that: In S3, within the transmission radius of each cluster head, with the goal of minimizing the weighted sum of the energy consumption of the drone and the average information age of the sensor nodes, an optimization problem is established and solved to obtain the best collection point for each cluster. The specific content is: In order to enable the UAV to complete the collection task as quickly as possible with the minimum energy consumption, the weighted sum of network information age and UAV energy consumption is minimized as the optimization goal; the average information age of the entire wireless network in n time slots is expressed as: in, is the total number of sensor nodes in the network; Establish the energy consumption model of the UAV; the propulsion power P of the UAV flying horizontally vH (v H (n)) is: where v H (n) is the horizontal flight speed of the UAV at time slot n, N u is the number of rotors of the UAV, c l is the rotor length of the UAV, ω u is the angular velocity, R u is the rotor disk radius, A u is the reference area of ​​the front end of the UAV, W u is the weight of the drone, ρ o and are air density and drag coefficient, λ u is a constant related to the physical quantity of the drone and the air density; Assume that at time slot n, the drone is in a hovering state and the horizontal speed of the drone is v H (n)=0, then the hovering power of the drone is P H for: Affected by air resistance and gravity, the propulsion power required for the drone to ascend and descend in the vertical direction is different; at time slot n, the vertical flight speed of the drone is v V (n), then the rising propulsion power and descent propulsion power The are: The transmission energy consumption of the drone is negligible; in time slot n, the energy consumption model of the drone is: Define δ1 and δ2 to represent the weights of the average information age of the wireless network and the energy consumption of the drone, respectively. Then the objective function is expressed as the weighted sum of the energy consumption of the drone and the network information age, that is: Assume that the cluster head acquisition strategy is G = {g1(n),g2(n),...,g M (n)},g i (n)∈{0,1} represents the indicator function of whether the UAV collects cluster head i data in time slot n. When g i When (n) = 0, it means that the UAV does not collect cluster head i. i When (n) = 1, the drone collects cluster head i; considering the drone 3D trajectory S and cluster head collection strategy G, for any positive integer N, the following optimization problem is established: Among them, constraints C1, C2, and C3 respectively represent restrictions on the UAV's flight space; constraint C4 divides the priority of clusters; constraint C5 limits the elevation angle range between the UAV and the cluster head; constraints C6 and C7 respectively limit whether the UAV is scheduled and the scheduling rules; constraint C8 limits the maximum transmission radius of the cluster head.

5. The method for designing a dynamic 3D trajectory of a UAV suitable for disaster scenarios according to claim 4, characterized in that: The S3 is specifically: S31: The UAV determines whether the UAV is within the maximum transmission radius based on the relative position of the current coordinates and the cluster head; if it is not within the maximum transmission radius, the coordinates of the UAV are updated; if it is within the maximum transmission radius, the following steps are continued in this step; S32: judging whether the current position is the hovering collection point of the cluster head according to the optimization objective function value, if not, continuing to update the coordinates of the drone; if it is a hovering collection point, continuing the following steps of this step; S33: The UAV sends a collection command, and the cluster head sends data information to the UAV to update the cluster information age.

6. The method for designing a dynamic 3D trajectory of a UAV suitable for disaster scenarios according to claim 5, characterized in that: In S32, within the maximum transmission radius of each cluster head, the dual deep Q network algorithm DDQN is used to calculate the corresponding cumulative objective function values ​​of all possible actions, and the collection strategy and the best hovering point when the minimum value is selected from all cluster heads according to the ε-greedy strategy. The best hovering point is the hovering collection point of the drone in the next time slot.

Images