Sensor perception data collection method based on cargo drone

Abstract

The present invention relates to the field of sensor data collection technology, and specifically to a sensor sensor data collection method based on a cargo drone, including: establishing a corresponding drone planning objective function with the goal of maximizing the amount of sensor data while satisfying the constraints of drone battery capacity, delivery delay, and data collection cost budget; solving the drone planning objective function to obtain a matching solution for the drone selected for collecting sensor data and its delivery route, and the data collection time allocated to the sensors covered by the corresponding delivery route; the selected drone delivers goods based on its delivery route, and collects sensor data from the sensors covered by its delivery route based on the corresponding data collection time; the selected drone uploads the sensor data it collects. The present invention can fully consider the operating status and mission requirements of the cargo drone, and can maximize the amount of sensor data collected.

Description

Technical Field

[0001] The present invention relates to the technical field of smart city data collection, and in particular to a sensor perception data collection method based on a cargo drone. Background Art

[0002] In recent years, the development of smart cities has been very rapid, thanks to the timely acquisition of various urban information. For this reason, more and more sensor networks have been widely deployed to monitor and collect various urban data. Timely and effective collection of these collected data and transmission to the data center for analysis and processing is a key link in ensuring the development of smart cities. The traditional data collection method is mainly through sensor networking, allowing data to reach the data center after multi-hop propagation in the network. However, this method has obvious defects. For example, since each sensor in the network not only needs to transmit its own data, but also acts as a relay to transmit data from other sensors, the battery capacity of the sensor itself is very small. The power will be quickly consumed, thus affecting the normal operation of the entire sensor network.

[0003] To this end, a lot of work has begun to focus on new data collection methods, among which the most extensive research is on the use of drones for sensing data collection. Drones have advantages such as agility and line-of-sight communication, which enable them to achieve low latency and high-efficiency data collection.

[0004] In addition, the rapid development of e-commerce in recent years has brought new challenges to the urban logistics industry. As a new type of vehicle, drones have unique advantages in the logistics and distribution process, which can effectively improve the distribution efficiency and greatly alleviate the pressure of urban logistics and transportation. Many large companies, such as Amazon, UPS, and JD.com, are considering investing drones in logistics and distribution. As a new logistics tool, cargo drones will gradually come into the public eye and become an indispensable member of smart cities. Based on this, the applicant thought of using cargo drones instead of traditional drones to collect sensor perception data, and reuse existing resources to greatly reduce the collection cost.

[0005] However, the applicant found that due to the particularity of cargo drones, the way they collect perception data is significantly different from that of traditional drones, which means that this work faces a new challenge. Specifically, the basic task of cargo drones is to deliver goods from warehouses to service stations. Each cargo has a certain weight, and different weights have different effects on the power of the drone, which directly affects the flight and hovering time of the drone; and each cargo has a certain delivery delay requirement, which also has a direct impact on the available time of the drone. At the same time, the lengths of each delivery route are different, and the number of sensors covered is also different. How to collect as much perception data as possible within a limited cost budget is also a technical problem. Therefore, how to design a data collection method that can take into account both the cargo delivery effect of cargo drones and the data collection effect of sensor network perception data is a technical problem that needs to be solved urgently. Summary of the invention

[0006] In view of the above-mentioned deficiencies in the prior art, the technical problem to be solved by the present invention is: how to provide a sensor perception data collection method based on cargo drones, so as to fully consider the operating status and mission requirements of the cargo drones, and to maximize the amount of perception data collected, thereby taking into account both the cargo delivery effect of the cargo drones and the data collection effect of the sensor network perception data, and further improving the efficiency and convenience of smart city sensor network perception data collection and uploading.

[0007] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0008] The sensor perception data collection method based on the cargo drone includes the following steps:

[0009] S1: Obtain a set of drones carrying goods, a set of optional delivery routes for the drones to choose from, and a set of sensors covered by the optional delivery routes;

[0010] S2: Under the constraints of drone battery capacity, cargo delivery delay and data collection cost budget, the corresponding drone planning objective function is established with the goal of maximizing the amount of perception data;

[0011] S3: Solve the drone planning objective function to obtain the matching scheme of the drones selected for collecting perception data and their delivery routes, as well as the data collection time allocated to the sensors covered by the corresponding delivery routes;

[0012] S4: Each selected drone delivers goods based on its delivery route and collects perception data from sensors covered by its delivery route based on the corresponding data collection time;

[0013] S5: Each selected drone uploads its collected perception data to complete the collection of sensor network perception data.

[0014] Preferably, in step S2, the drone planning objective function is expressed by the following formula:

[0015]

[0016]

[0017]

[0018]

[0019]

[0020]

[0021] Where: U represents the amount of sensory data collected; x and t are both decision variables; x = {x ij} represents the matching scheme of drones and their delivery routes, that is, whether drone i chooses delivery route j, x ij =1 means yes, x ij =0 means no; t = {t jk} represents the data collection time assigned to the sensor; u jk (t jk ) represents the data collection time t jk The amount of sensory data collected within; d j represents the route length of delivery route j; v represents the flight speed of the drone; c ij Indicates that the weight of drone i is m i The cost of collecting sensor data for goods on delivery route j; represents a collection of drones; Represents a set of optional delivery routes; represents the sensor set covered by the delivery route j; E u 、T i , C represent the upper limits of the drone battery capacity, delivery delay, and data collection cost budget, respectively.

[0022] Preferably, the amount of perception data collection u is calculated by the following formula: jk (t jk );

[0023]

[0024] Where: B represents the bandwidth when the drone and the sensor communicate; β represents the signal-to-noise ratio when the reference distance is one meter; Indicates sensorjk The remaining power; H represents the communication distance between the UAV and the sensor, that is, the flight altitude of the UAV; α represents the path loss parameter; Indicates sensor jk The amount of perception data stored.

[0025] Preferably, the cost c is calculated by the following formula: ij ;

[0026]

[0027] in,

[0028]

[0029] Where: c0 represents the maintenance cost of a single drone; ρ represents the electricity cost per unit of electricity; e ij Indicates that the weight of drone i is m i The total energy consumption of collecting sensor data for goods on delivery route j; Indicates that the weight of drone i is m i The basic energy consumption when transporting goods, that is, the energy consumption when flying directly from the warehouse to the service station without collecting data; P f (m i ) and P h (m i ) represent the weight of UAV i carried by m i The flight power and hovering power when carrying cargo; v represents the flight speed of the UAV; d j represents the route length of delivery route j; d0 represents the length of the shortest route from the warehouse to the service station.

[0030] Preferably, in step S3, the drone planning objective function is solved by the following steps:

[0031] S301: Assuming that the decision variable x is known, calculate the available power converted from the remaining cost budget after deducting the maintenance cost of the drone and the flight energy consumption electricity cost, that is, the remaining energy E that all drones can use for hovering r ;

[0032] S302: Remaining energy E r Perform sharding processing, and the size of each energy slice is e0;

[0033] S303: Allocate energy slices based on a greedy strategy, that is, for each given decision variable x, the data collection time allocated to the sensor based on the decision variable x can be obtained through the greedy strategy, so that the decision variable x and t form a mapping relationship, and an approximate proxy model containing only the decision variable x is obtained;

[0034] S304: converting the approximate proxy model into an equivalent set function form to obtain a corresponding simplified objective function;

[0035] S305: Solve the simplified objective function based on the local search matching search algorithm to obtain the matching solution of the drone and its delivery route, as well as the data collection time allocated to the sensors covered by the corresponding delivery route.

[0036] Preferably, in step S301, the remaining energy E of all drones available for hovering is calculated by the following steps: r ;

[0037]

[0038] Where: B represents the upper limit of the data collection cost budget; ρ represents the electricity cost per unit of electricity; c0 represents the maintenance cost of a single drone; P f (m i ) indicates that the weight of the drone is m i The flight power when the cargo is loaded; v represents the flight speed of the drone; d j represents the route length of delivery route j; d0 represents the length of the shortest route from the warehouse to the service station.

[0039] Preferably, in step S302, the number of remaining energy slices can be expressed as Allocate energy slices to sensors slice by slice;

[0040] The sensor s is expressed by the following formula jk The benefit obtained when the zth energy tablet is distributed

[0041]

[0042] in,

[0043] Where: represents the weight of the goods delivered on the delivery route j; e0 represents the size of the energy tablet; P h (m i ) indicates that the weight of the vehicle on delivery route j is Hovering power when the cargo is Indicates sensor jk The amount of perception data collected when a total of z energy slices are allocated, ze0 represents the allocated energy; Indicates sensor jk The amount of sensory data collected when a total of z-1 energy slices are allocated; Indicates sensor jk The amount of perception data collected when the first energy slice is allocated.

[0044] Preferably, in step S303, a total of Γ energy slices are iteratively allocated. Each time an energy slice is allocated, the following formula is used to greedily select the energy slice that maximizes the current energy slice benefit and satisfies the drone battery capacity E u and delivery delay T i The sensors are allocated for data collection time;

[0045]

[0046] Where: represents the sensor selected each time the energy slice is allocated; S represents the set of all sensors on the selected route; Indicates that at the nth allocation, s jk The number of energy slices that have been allocated; Indicates that in the nth allocation, sensor s jk Redistribute the benefits of one energy tablet;

[0047] When no sensor meets the allocation requirements or the energy slices are allocated, the allocation is terminated;

[0048] The sensor s is calculated by combining the allocated energy slices using the following formula: jk the time allocated for data collection;

[0049]

[0050] Where: t jk Indicates sensor jk Allotted time for data collection; P h (m i ) represents the load on delivery route j Hover power at 100 Hz;

[0051] The approximate proxy model containing only the decision variable x is expressed by the following formula;

[0052]

[0053]

[0054]

[0055] Where: t jk (x) indicates that when a decision variable x is given, sensor s jk The data collection time allocated based on the allocation strategy; jk (t jk (x)) represents sensor s jk According to the allocated data collection time t jk (x) The amount of sensory data collected.

[0056] Preferably, in step S304, Represents the combination of all drones and delivery routes;

[0057] use represents the set of selected drone and delivery route combinations,

[0058] based on Calculate the objective function

[0059] Based on the objective function Generate the following simplified objective function;

[0060]

[0061]

[0062]

[0063] Where: is an indicator function, which means that if Then the function value is +1.

[0064] Preferably, the matching search algorithm based on local search includes two levels of iteration: inner iteration and outer iteration;

[0065] Where n∈{1,...,N °} represents the index of the outer iteration; and Represent the candidate matching set and the local optimal matching set in the nth iteration respectively;

[0066] 1) Inner iteration, in the candidate matching set Search for a local optimal matching set in

[0067] First, use a match with the maximum benefit initialization

[0068] Then two local search-based operations are used to iteratively search for new matching sets with higher benefits in the neighborhood, including deletion and exchange operations. These two operations are iteratively performed to search for matching sets with higher benefits until a local optimal solution is obtained. Among them, the deletion operation, that is, from Delete an edge Exchange operation, ready to use The two edges in Exchange does not belong to An edge of

[0069] 2) Outer iteration: in the local optimal matching set found Select the best local matching set

[0070] First use initialization

[0071] Then, in each iteration n, an internal iteration is performed to search for a local optimal match.

[0072] Afterwards, by removing the found matching sets , update the candidate matching set, that is, Feedback to search for another local optimal match in the next iteration n+1

[0073] Finally, at the end of the outer iteration, the local optimal matching set with the maximum benefit is selected As a final result;

[0074] Based on the best match Combine the following formulas to calculate the matching scheme x, data collection time t, and perception data volume U of the drone and its delivery route respectively;

[0075]

[0076]

[0077]

[0078] The sensor perception data collection method based on the cargo drone in the present invention has the following beneficial effects:

[0079] The present invention selects drones for collecting data and assigns corresponding delivery routes to them, and at the same time assigns data collection times to sensors covered by the delivery routes, so that each selected drone can deliver goods based on its delivery route and collect perception data of sensors covered by its delivery route based on the corresponding data collection time. That is, it is possible to effectively collect sensor perception data through cargo drones, thereby improving the efficiency and convenience of collecting and uploading perception data of smart city sensor networks.

[0080] At the same time, the present invention establishes a corresponding drone planning objective function with the goal of maximizing the amount of perceived data while satisfying the constraints of drone battery capacity, delivery delay and data collection cost budget, so that it is possible to more reasonably select drones and their delivery routes for collecting data, and at the same time better allocate data collection time to sensors covered by the delivery routes. That is, on the premise of completing the sensor perception data collection, it is possible to fully consider the operating status and mission requirements of the cargo drone and maximize the amount of perception data collected, thereby taking into account both the cargo delivery effect of the cargo drone and the data collection effect of the sensor network perception data, and further improving the efficiency and convenience of smart city sensor network perception data collection and uploading. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] In order to make the purpose, technical solution and advantages of the invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings, in which:

[0082] Figure 1 A logical block diagram of the sensor perception data collection method based on cargo drones. DETAILED DESCRIPTION

[0083] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but only represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the present invention.

[0084] It should be noted that similar numbers and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. In the description of the present invention, it should be noted that the orientation or position relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside", etc. is based on the orientation or position relationship shown in the drawings, or the orientation or position relationship in which the invention product is usually placed when used, 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 operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention. In addition, the terms "first", "second", "third", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance. In addition, the terms "horizontal", "vertical", etc. do not mean that the components are absolutely horizontal or suspended, but can be slightly tilted. For example, "horizontal" only means that its direction is more horizontal than "vertical", and does not mean that the structure must be completely horizontal, but can be slightly tilted. In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0085] Example:

[0086] This embodiment discloses a method for collecting sensor perception data based on a cargo drone.

[0087] like Figure 1 As shown, the sensor perception data collection method based on the cargo drone includes the following steps:

[0088] S1: Obtain a set of drones carrying goods, a set of optional delivery routes for the drones to choose from, and a set of sensors covered by the optional delivery routes;

[0089] S2: Under the constraints of drone battery capacity, cargo delivery delay and data collection cost budget, the corresponding drone planning objective function is established with the goal of maximizing the amount of perception data;

[0090] S3: Solve the drone planning objective function to obtain the matching scheme of the drones selected for collecting perception data and their delivery routes, as well as the data collection time allocated to the sensors covered by the corresponding delivery routes;

[0091] S4: Each selected drone delivers goods based on its delivery route and collects perception data from sensors covered by its delivery route based on the corresponding data collection time;

[0092] S5: Each selected drone uploads its collected perception data to complete the collection of sensor network perception data.

[0093] The present invention selects drones for collecting data and assigns corresponding delivery routes to them, and at the same time assigns data collection times to sensors covered by the delivery routes, so that each selected drone can deliver goods based on its delivery route and collect perception data of sensors covered by its delivery route based on the corresponding data collection time. That is, it is possible to effectively collect sensor perception data through cargo drones, thereby improving the efficiency and convenience of collecting and uploading perception data of smart city sensor networks.

[0094] At the same time, the present invention establishes a corresponding drone planning objective function with the goal of maximizing the amount of perceived data while satisfying the constraints of drone battery capacity, delivery delay and data collection cost budget, so that it is possible to more reasonably select drones and their delivery routes for collecting data, and at the same time better allocate data collection time to sensors covered by the delivery routes. That is, on the premise of completing the sensor perception data collection, it is possible to fully consider the operating status and mission requirements of the cargo drone and maximize the amount of perception data collected, thereby taking into account both the cargo delivery effect of the cargo drone and the data collection effect of the sensor network perception data, and further improving the efficiency and convenience of smart city sensor network perception data collection and uploading.

[0095] In this embodiment, it is assumed that a freight company has I drones for delivering goods between a warehouse and a service station. The distance between the warehouse and the service station is denoted by d0, and the drone collection is denoted by Each drone has a corresponding cargo to be delivered. Different cargoes have different weight and delivery delay requirements. i and T i The set of optional delivery routes of drones from warehouses to service stations is represented by To indicate, there are J items in total, use d j Indicates the length of the jth delivery route. A large number of sensors for urban monitoring and perception are deployed in the area from the warehouse to the service station. Each route will be covered by a certain number of sensors. The sensors covered by route j are To represent, where s jk represents the kth sensor on route j, sensor s jk The remaining power and the allocated sensing collection time are respectively and t jk To express.

[0096] When collecting sensor data, the drone carries cargo and flies to the air near the sensor and hovers. After receiving the data transmission command from the drone above it, the sensor transmits the stored sensor data to the drone through wireless communication technology, and then the drone performs further data transfer. Since the process without sending instructions is very short, its corresponding time can be ignored. Due to the mobile characteristics of the drone, its communication process with the sensor can be regarded as line-of-sight communication, which is mainly affected by the communication distance between the two.

[0097] Assuming the drone flies at a fixed altitude, the sensor s jk The data transmission rate can be expressed as:

[0098]

[0099] Among them, B represents the bandwidth when the two communicate; α and β represent the path loss parameter and the signal-to-noise ratio when the reference distance is one meter, respectively; H represents the communication distance between the drone and the sensor, that is, the flight altitude of the drone; Indicates sensor jk The data transmission power can be adjusted within a certain range.

[0100] The drone at time t jk Internal slave sensors jk The collected time can be expressed as:

[0101]

[0102] Because the sensors jk The battery power is limited, and its data transmission power meets In addition, each sensor only stores a certain amount of sensing data. Indicates sensor jk The amount of perception data stored.

[0103] The amount of perception data collection u is calculated by the following formula jk (t jk );

[0104]

[0105] Where: B represents the bandwidth when the drone and the sensor communicate; β represents the signal-to-noise ratio when the reference distance is one meter; Indicates sensor jk The remaining power; H represents the communication distance between the UAV and the sensor, that is, the flight altitude of the UAV; α represents the path loss parameter; H αIt means that in wireless communication, there is a certain relationship between data transmission rate and transmission distance; Indicates sensor jk The amount of perception data stored.

[0106] The cargo drone collects sensor perception data during the delivery process, and its energy consumption is mainly from flying and hovering. In addition, because the flight and hovering power of the drone are related to the weight of the cargo it carries, and the power increases linearly with the increase of cargo weight, P f (·) and P (h) (·) represents the flight and hovering power function of the UAV, then the weight of the cargo UAV is m i The total energy consumption of the sensors on route j when collecting goods is:

[0107]

[0108] Among them, P f (m i ) and P h (m i ) represent the load in m i The flight power and hovering power of the drone.

[0109] It costs a certain amount of extra money for a cargo drone to collect sensor perception data. The cost mainly includes two parts: one is the maintenance cost of the drone after each data collection, c0 is used to represent the maintenance cost of a single drone; the other is the electricity cost used by the drone to collect data, ρ is used to represent the electricity cost per unit of electricity.

[0110] The cost c is calculated by the following formula ij ;

[0111]

[0112] in,

[0113]

[0114] Where: c0 represents the maintenance cost of a single drone; ρ represents the electricity cost per unit of electricity; e ij Indicates that the weight of drone i is m i The total energy consumption of collecting sensor data for goods on delivery route j; Indicates that the weight of drone i is m i The basic energy consumption when transporting goods, that is, the energy consumption when flying directly from the warehouse to the service station without collecting data; P f (m i ) and P h (m i ) represent the weight of UAV i carried by mi The flight power and hovering power when carrying cargo; v represents the flight speed of the UAV; d j represents the route length of delivery route j; d0 represents the length of the shortest route from the warehouse to the service station.

[0115] The present invention proposes a perception data collection problem based on cargo drones (DIDC problem), namely: given I drones carrying corresponding delivery goods and J optional delivery routes, u , delivery delay T i Under the premise of the constraints of the data collection cost budget B, the corresponding delivery route j is selected for drone i, and the sensors covered on the delivery route j are given {s jk}Assign corresponding data collection time {t jk}, so as to maximize the total amount of perception data collected.

[0116] The drone planning objective function is expressed by the following formula:

[0117]

[0118] (Delivery delay constraints for goods)

[0119] (Energy Consumption Constraints of Drones)

[0120] (Data collection cost constraints)

[0121] (The selection constraints between drones and routes are that each delivery route can only be selected by at most one drone at a time, and each drone can only select at most one delivery route at a time).

[0122] Where: U represents the amount of sensory data collected; x and t are both decision variables; x = {x ij} represents the matching scheme of drones and their delivery routes, that is, whether drone i chooses delivery route j, x ij =1 means yes, x ij =0 means no; t = {t jk} represents the data collection time assigned to the sensor; u jk (t jk ) represents the data collection time t jk The amount of sensory data collected within; d j represents the route length of delivery route j; v represents the flight speed of the drone; c ij Indicates that the weight of drone i is m i The cost of collecting sensor data for goods on delivery route j; represents a collection of drones; Represents a set of optional delivery routes; represents the sensor set covered by the delivery route j; E u , T i , C represent the upper limits of the drone battery capacity, delivery delay, and data collection cost budget, respectively.

[0123] The above-mentioned drone planning objective function established by the present invention can maximize the amount of perceived data while satisfying the constraints of drone battery capacity, delivery delay and data collection cost budget, so that the drone and its delivery route for collecting data can be selected more reasonably, and at the same time, the data collection time allocated to the sensors covered by the delivery route can be better allocated. That is, on the premise of completing the sensor perception data collection, the operating status and mission requirements of the cargo drone can be fully considered, and the amount of perception data collection can be maximized, so as to better balance the cargo delivery effect of the cargo drone and the data collection effect of the sensor network perception data.

[0124] In the specific implementation process, the drone planning objective function is solved through the following steps:

[0125] S301: Assuming that the decision variable x is known, calculate the available power converted from the remaining cost budget after deducting the maintenance cost of the drone and the flight energy consumption electricity cost, that is, the remaining energy E that all drones can use for hovering r ; Because every time a drone is selected to collect data, there will be corresponding maintenance costs for the drone and electricity costs for the drone to fly along the route. After deducting these two costs, the remaining cost budget is used for hovering and collecting data.

[0126] S302: Remaining energy E r Perform sharding processing, and the size of each energy slice is e0;

[0127] S303: Allocate energy slices based on a greedy strategy, that is, for each given decision variable x, the data collection time allocated to the sensor based on the decision variable x can be obtained through the greedy strategy, so that the decision variables x and t form a mapping relationship, and an approximate proxy model containing only the decision variable x is obtained; at this time, only x needs to be optimized.

[0128] S304: converting the approximate proxy model into an equivalent set function form to obtain a corresponding simplified objective function;

[0129] S305: Solve the simplified objective function based on the local search matching search algorithm to obtain the matching solution of the drone and its delivery route, as well as the data collection time allocated to the sensors covered by the corresponding delivery route.

[0130] The present invention assumes that the decision variables are known, calculates the remaining energy of all drones, and slices the remaining energy of all drones, and then distributes the energy slices based on a greedy strategy to obtain an approximate proxy model containing only the decision variables, and converts the approximate proxy model into an equivalent set function form, so that an approximate proxy model can be constructed based on energy slicing to approximately convert the drone planning objective function into a non-negative, non-monotonic, classical optimization problem of a submodule function with two pseudo-matrix constraints, and further solves the optimization problem through a local search matching search algorithm, so as to obtain an optimal solution with higher performance, that is, it is possible to more reasonably select drones and their distribution routes for collecting data, and at the same time, better allocate data collection time to sensors covered by the distribution routes, and then, on the premise of completing the sensor perception data collection, fully consider the operating status and task requirements of the cargo drone, and maximize the amount of perception data collection, so as to better balance the cargo distribution effect of the cargo drone and the data collection effect of the sensor network perception data.

[0131] In the specific implementation process, when the value of variable x is given, the remaining energy E can be calculated r ,This energy consumption represents the remaining energy available for hovering after the UAV maintenance costs and the electricity costs for flight.

[0132] The remaining energy E available for hovering of all drones is calculated by the following steps: r :

[0133]

[0134] Where: C represents the upper limit of the data collection cost budget; ρ represents the electricity cost per unit of electricity; c0 represents the maintenance cost of a single drone; P f (m i ) indicates that the weight of the drone is m i The flight power when the cargo is loaded; v represents the flight speed of the drone; d j represents the route length of delivery route j; d0 represents the length of the shortest route from the warehouse to the service station.

[0135] The number of remaining energy slices can be expressed as Allocate energy slices to sensors slice by slice;

[0136] The sensor s is expressed by the following formula jk The benefit obtained when the zth energy tablet is distributed

[0137]

[0138] in,

[0139] Where: represents the weight of the goods delivered on the delivery route j; e0 represents the size of the energy tablet; P h (m i ) indicates that the weight of the vehicle on delivery route j is Hovering power when the cargo is Indicates sensor jk The amount of perception data collected when a total of z energy slices are allocated, ze0 represents the energy allocated, Energy divided by power represents the time allocated, and substituting it into the previous objective function, we get the benefit; Indicates sensor jk The amount of sensory data collected when a total of z-1 energy slices are allocated is the benefit that can be obtained by subtracting the latter from the former; Indicates sensor jk The amount of perception data collected when the first energy slice is allocated.

[0140] In the specific implementation process, a total of Γ energy slices are iteratively allocated. Each time an energy slice is allocated, the following formula is used to greedily select the energy slice that maximizes the current energy slice benefit and satisfies the UAV battery capacity E u and delivery delay T i The sensors are allocated for data collection time;

[0141]

[0142] Where: represents the sensor selected each time the energy slice is allocated; S represents the set of all sensors on the selected route; Indicates that at the nth allocation, s jk The number of energy slices that have been allocated; Indicates that in the nth allocation, sensor s jk Redistribute the benefits of one energy tablet;

[0143] When no sensor meets the allocation requirements or the energy slices are allocated, the allocation is terminated;

[0144] The sensor s is calculated by combining the allocated energy slices using the following formula: jk the time allocated for data collection;

[0145]

[0146] Where: t jk Indicates sensor jk Allotted time for data collection; P h (m i ) represents the load on delivery route j Hover power at 100 Hz;

[0147] In this embodiment, the energy slice allocation algorithm based on the greedy strategy is as follows:

[0148]

[0149]

[0150] For any given value of decision variable x, the time allocation plan t corresponding to x can be obtained through the above scheme.

[0151] The approximate proxy model containing only the decision variable x is expressed by the following formula;

[0152]

[0153]

[0154]

[0155] Where: t jk (x) indicates that when a decision variable x is given, sensor s jk The data collection time allocated based on the allocation strategy; jk (t jk (x)) represents sensor s jk According to the allocated data collection time t jk (x) The amount of sensory data collected.

[0156] It should be noted that the approximate proxy model constructed in the present invention has a 1 / 2 performance guarantee, and the specific proof is shown in Theorem 1.

[0157] Theorem 1, the performance loss of the approximate proxy model does not exceed 1 / 2, that is,

[0158] prove:

[0159] Let (x * ,t * ) represents the optimal solution to the DIDC problem, that is,

[0160]

[0161] When the drone is in t jk During each s jk When collecting sensor data, the corresponding hovering energy consumption is

[0162] Regarding the energy slice unit e0, we let and Respectively represent The integer part and the remainder of .

[0163] therefore, can be decomposed into two parts, namely

[0164]

[0165] in

[0166] Defining a function because F(ζ) is a concave function. Similar to F(ζ), the function u jk (t jk ) is also concave. Due to the subadditive property of concave functions, according to We have

[0167] In addition, F′(ζ) is a function that decreases with ζ, because F″(ζ)<0. Therefore, Therefore, F(ζ) is a non-decreasing function, ζ. Similarly, u jk (t jk ) is also a random jk A decreasing function. At the same time,

[0168] thus,

[0169] Therefore, based on We have

[0170] Furthermore,

[0171] When t is a discrete decision variable based on energy partitioning. ) can be equivalently transformed into a 0-1 knapsack problem, where the optimal solution can be achieved by using a greedy solution.

[0172] Since the algorithm uses a greedy sharding allocation scheme to construct an approximate proxy model, F(x * ) is the optimal solution to this problem, and and is a feasible solution. Therefore, we have

[0173]

[0174]

[0175] because Can get

[0176]

[0177]

[0178] Therefore, Theorem 1 is proved. The proof is complete.

[0179] In order to facilitate the analysis of function properties, the objective function is further transformed into an equivalent set function form.

[0180] use Represents the combination of all drones and delivery routes;

[0181] use represents the set of selected drone and delivery route combinations,

[0182] based on Calculate the objective function

[0183] Based on the objective function Generate the following simplified objective function;

[0184]

[0185]

[0186]

[0187] Where: is an indicator function, which means that if Then the function value +1, the formula means that for any i and j, the number of matches must be less than or equal to 1, which means that In the example, any i can only match at most one j, which is equivalent to It represents the objective function after being converted into a set function, which is equivalent to F(x), except that the x variable in F(x) is converted into a set variable.

[0188] It should be noted that the simplified objective function after transformation in the present invention is to maximize a non-negative, non-monotonic submodular function with two pseudo-matrix constraints. See Theorem 2 for specific proof.

[0189] Theorem 2, in the constructed set function model, the objective function is non-negative, non-monotonic, submodular, and constrained to two pseudo-matrix constraints.

[0190] prove:

[0191] i) Non-negativity. Obviously, according to the profit calculation model, the objective function is non-negative.

[0192] ii) Non-monotonicity. Given when hour, and is to Add enough elements to The remaining energy E r ≤0, so we can get so So the objective function is non-monotonic.

[0193] iii) Submodularity. We should assume that The following relationship holds,

[0194]

[0195] Let t (0) ,t (1) ,t (2) and t (3) Indicates that according to Alg. and The following data collection time is allocated. jk (t jk ) represents the transmission time t of sensor k jk Therefore, we can express the objective function as

[0196]

[0197]

[0198]

[0199]

[0200] Based on the above, we have

[0201]

[0202]

[0203] The first term indicates The reward for the task is reduced and consists of two parts. The additional flight energy cost and reuse cost caused by R1 reduce the total energy budget, thereby reducing The effectiveness of the task. The new time allocation for task r1 reduces this utility. The second term represents the increase in utility for task r1, expressed as

[0204] therefore,

[0205] Similarly,

[0206] Since we use a greedy energy slice allocation strategy based on the maximum amount of sensing data that each energy slice can obtain, and Can get, and

[0207] Therefore, H(·) is submodular.

[0208] A pair of constructed Satisfy the three conditions of the matroid:

[0209] i) According to the formula It is easy to prove therefore,

[0210] ii) According to the formula because therefore, Not true, the second condition is true.

[0211] iii) Due to Existence by Selected goods But not Select.

[0212]

[0213] therefore, The third condition is met.

[0214] Therefore, Theorem 2 is proved. Proof completed.

[0215] In the specific implementation process, the matching search algorithm based on local search includes two levels of iteration: inner iteration and outer iteration;

[0216] Where n∈{1,...,N°} represents the index of the outer iteration; and Represent the candidate matching set and the local optimal matching set in the nth iteration respectively;

[0217] 1) Inner iteration, in the candidate matching set Search for a local optimal matching set in

[0218] First, use a match with the maximum benefit initialization

[0219] Then two local search-based operations are used to iteratively search for new matching sets with higher benefits in the neighborhood, including deletion and exchange operations. These two operations are iteratively performed to search for matching sets with higher benefits until a local optimal solution is obtained. Among them, the deletion operation, that is, from Delete an edge Swap operation, ready to use The two edges in Exchange does not belong to An edge of

[0220] 2) Outer iteration: in the local optimal matching set found Select the best local matching set

[0221] First use initialization

[0222] Then, in each iteration n, an internal iteration is performed to search for a local optimal match.

[0223] Afterwards, by removing the found matching sets , update the candidate matching set, that is, Feedback to search for another local optimal match in the next iteration n+1

[0224] Finally, at the end of the outer iteration, the local optimal matching set with the maximum benefit is selected As a final result;

[0225] Based on the best match Combine the following formulas to calculate the matching scheme x, data collection time t, and perception data volume U of the drone and its delivery route respectively;

[0226]

[0227]

[0228]

[0229] In this embodiment, the drone route matching algorithm based on local search is as follows.

[0230]

[0231]

[0232] 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 the technical solution. Those skilled in the art should understand that those modifications or equivalent substitutions of the technical solution of the present invention that do not depart from the purpose and scope of the technical solution should be included in the scope of the claims of the present invention.

Claims

1. A sensor perception data collection method based on a cargo drone, characterized in that: The following steps are involved: S1: Obtain a set of drones carrying goods, a set of optional delivery routes for the drones to choose from, and a set of sensors covered by the optional delivery routes; S2: Under the constraints of drone battery capacity, cargo delivery delay and data collection cost budget, the corresponding drone planning objective function is established with the goal of maximizing the amount of perception data; In step S2, the drone planning objective function is expressed by the following formula: Where: U represents the amount of sensory data collected; x and t are both decision variables; x = {x ij } represents the matching scheme of drones and their delivery routes, that is, whether drone i chooses delivery route j, x ij =1 means yes, x ij =0 means no; t = {t jk } represents the data collection time assigned to the sensor; u jk (t jk ) represents the data collection time t jk The amount of sensory data collected within; d j represents the route length of delivery route j; v represents the flight speed of the drone; c ij Indicates that the weight of drone i is m i The cost of collecting sensor data for goods on delivery route j; represents a collection of drones; Represents a set of optional delivery routes; represents the sensor set covered by the delivery route j; E u 、T i , C represent the upper limits of the drone battery capacity, delivery delay, and data collection cost budget respectively; The amount of perception data collection u is calculated by the following formula jk (t jk ); Where: B represents the bandwidth when the drone and the sensor communicate; β represents the signal-to-noise ratio when the reference distance is one meter; Indicates sensor jk The remaining power; H represents the communication distance between the UAV and the sensor, that is, the flight altitude of the UAV; α represents the path loss parameter; Indicates sensor jk The amount of sensory data stored; The cost c is calculated by the following formula ij ; in, Where: c0 represents the maintenance cost of a single drone; ρ represents the electricity cost per unit of electricity; e ij Indicates that the weight of drone i is m i The total energy consumption of collecting sensor data for goods on delivery route j; Indicates that the weight of drone i is m i The basic energy consumption when transporting goods, that is, the energy consumption when flying directly from the warehouse to the service station without collecting data; P f (m i ) and P h (m i ) represent the weight of UAV i carried by m i The flight power and hovering power when carrying cargo; v represents the flight speed of the UAV; d j represents the route length of delivery route j; d0 represents the length of the shortest route from the warehouse to the service station; S3: Solve the drone planning objective function to obtain the matching scheme of the drones selected for collecting perception data and their delivery routes, as well as the data collection time allocated to the sensors covered by the corresponding delivery routes; In step S3, the UAV planning objective function is solved by the following steps: S301: Assuming that the decision variable x is known, calculate the available power converted from the remaining cost budget after deducting the maintenance cost of the drone and the flight energy consumption electricity cost, that is, the remaining energy E that all drones can use for hovering r ; S302: Remaining energy E r Perform sharding processing, and the size of each energy slice is e0; S303: Allocate energy slices based on a greedy strategy, that is, for each given decision variable x, the data collection time allocated to the sensor based on the decision variable x can be obtained through the greedy strategy, so that the decision variable x and t form a mapping relationship, and an approximate proxy model containing only the decision variable x is obtained; S304: converting the approximate proxy model into an equivalent set function form to obtain a corresponding simplified objective function; S305: Solving the simplified objective function based on the matching search algorithm of the local search, obtaining the matching solution of the drone and its delivery route, and the data collection time allocated to the sensors covered by the corresponding delivery route; S4: Each selected drone delivers goods based on its delivery route and collects perception data from sensors covered by its delivery route based on the corresponding data collection time; S5: Each selected drone uploads its collected perception data to complete the collection of sensor network perception data.

2. The sensor perception data collection method based on the cargo drone according to claim 1, characterized in that: In step S301, the remaining energy E of all drones available for hovering is calculated by the following steps: r ; Where: C represents the upper limit of the data collection cost budget; ρ represents the electricity cost per unit of electricity; c0 represents the maintenance cost of a single drone; P f (m i ) indicates that the weight of the drone is m i The flight power when the cargo is loaded; v represents the flight speed of the drone; d j represents the route length of delivery route j; d0 represents the length of the shortest route from the warehouse to the service station.

3. The sensor perception data collection method based on the cargo drone as claimed in claim 2, characterized in that: In step S302, the number of remaining energy slices can be expressed as Allocate energy slices to sensors slice by slice; The sensor s is expressed by the following formula jk The benefit obtained when the zth energy tablet is distributed in, Where: represents the weight of the goods delivered on the delivery route j; e0 represents the size of the energy piece; Indicates that the weight of the delivery route j is Hovering power when the cargo is Indicates sensor jk The amount of perception data collected when a total of z energy slices are allocated, ze0 represents the allocated energy; Indicates sensor jk The amount of sensory data collected when a total of z-1 energy slices are allocated; Indicates sensor jk The amount of perception data collected when the first energy slice is allocated.

4. The method for collecting sensor perception data based on a cargo drone as claimed in claim 3, characterized in that: In step S303, a total of Γ energy slices are iteratively allocated. Each time an energy slice is allocated, the following formula is used to greedily select the energy slice that maximizes the current energy slice benefit and satisfies the drone battery capacity E u and delivery delay T i The sensors are allocated for data collection time; Where: represents the sensor selected each time the energy slice is allocated; S represents the set of all sensors on the selected route; Indicates that at the nth allocation, s jk The number of energy slices that have been allocated; Indicates that in the nth allocation, sensor s jk Redistribute the amount of sensory data that can be obtained by one energy slice; When no sensor meets the allocation requirements or the energy slices are allocated, the allocation is terminated; The sensor s is calculated by combining the allocated energy slices using the following formula: jk the time allocated for data collection; Where: t jk Indicates sensor jk the time allocated for data collection; Indicates the load on delivery route j Hover power at 100 Hz; The approximate proxy model containing only the decision variable x is expressed by the following formula; Where: t jk (x) indicates that when a decision variable x is given, sensor s jk The data collection time allocated based on the allocation strategy; u jk (t jk (x)) represents sensor s jk According to the allocated data collection time t jk (x) The amount of sensory data collected.

5. The method for collecting sensor perception data based on a cargo drone as claimed in claim 4, characterized in that: In step S304, Represents the combination of all drones and delivery routes; use represents the set of selected drone and delivery route combinations, based on Calculate the objective function Based on the objective function Generate the following simplified objective function; Where: is an indicator function, which means that if Then the function value is +1.

6. The method for collecting sensor perception data based on a cargo drone as claimed in claim 5, characterized in that: In step S305, the matching search algorithm based on local search includes two levels of iteration: inner iteration and outer iteration; Where n∈{1,...,N o } represents the index of the outer iteration; and Represent the candidate matching set and the local optimal matching set in the nth iteration respectively; 1) Inner iteration, in the candidate matching set Search for a local optimal matching set in First, use a match with the maximum benefit initialization Then two local search-based operations are used to iteratively search for new matching sets with higher benefits in the neighborhood, including deletion and exchange operations. These two operations are iteratively performed to search for matching sets with higher benefits until a local optimal solution is obtained. Among them, the deletion operation, that is, from Delete an edge Swap operation, ready to use The two edges in Exchange does not belong to An edge of 2) Outer iteration: in the local optimal matching set found Select the best local matching set First use initialization Then, in each iteration n, an internal iteration is performed to search for a local optimal match. Afterwards, by removing the found matching sets , update the candidate matching set, that is, Feedback to search for another local optimal match in the next iteration n+1 Finally, at the end of the outer iteration, the local optimal matching set with the maximum benefit is selected As a final result; Based on the best match Combine the following formulas to calculate the matching scheme x, data collection time t, and perception data volume U of the drone and its delivery route respectively;

Images