CN-122019011-A - Unmanned energy consumption optimization method based on end-side-cloud cooperative architecture and SHADE-NE algorithm

CN122019011ACN 122019011 ACN122019011 ACN 122019011ACN-122019011-A

Abstract

The invention belongs to the technical field of unmanned aerial vehicle systems. A unmanned aerial vehicle energy consumption optimization method based on an end-side-cloud cooperative framework and SHADE-NE algorithm is characterized by comprising the following steps of 1) adopting the end-side-cloud cooperative framework, wherein the framework firstly utilizes an unmanned aerial vehicle to complete real-time acquisition and preliminary processing of data by establishing a layered cooperative mechanism among an end side, an edge end and a cloud end of the unmanned aerial vehicle, and then judges a task unloading position through a management strategy to realize layered execution of task processing, and 2) adopting a SHADE-NE optimization algorithm to solve the task unloading and resource scheduling process of the unmanned aerial vehicle in an iterative search mode, so that an optimal solution can be obtained under the condition of meeting various constraint conditions. The method can improve the real-time performance and reliability of unmanned aerial vehicle task processing, enhance the collaborative energy consumption optimizing capability of multiple unmanned aerial vehicles, and improve the stability and adaptability of the optimizing process.

Inventors

YANG CHUYUAN
LIU SHIBIN
LU YUN
CHEN BIN
XIONG QI
LI XIANSHAN
LI XINYAN
GAN YAN
WANG FENG
FANG WEN
LI FEI
LI HUANGQIANG
XIE LIHUI
WANG XUWEI
WANG MIN
LUO CHAO
XIANG CHUAN
QIN SIYU
XIE QIONGYAO
XIANG KUN
YANG LINGXI
Zhu tianan
WANG LEI
XIANG YING
HE QI
WANG MENG
Duan Junge
LI HAO
XIA JUNRONG
SONG LEI
YAO JUNWEI
FAN LIPING
DENG LING
Ding Haotao
Tao Yinyu

Assignees

国网湖北省电力有限公司宜昌供电公司
宜昌长江三峡岸电技术有限公司
三峡大学

Dates

Publication Date: 20260512
Application Date: 20260203

Claims (2)

1. The unmanned energy consumption optimization method based on the end-side-cloud cooperative architecture and SHADE-NE algorithm is characterized by comprising the following steps: 1) The method comprises the steps of adopting an end-side-cloud cooperative framework, establishing a layered cooperative mechanism among an end side, an edge end and a cloud end of an unmanned aerial vehicle, firstly utilizing the unmanned aerial vehicle to complete real-time acquisition and preliminary processing of data, and judging the unloading position of a task through a management strategy to realize layered execution of task processing; 2) And adopting SHADE-NE optimization algorithm, solving in an iterative search mode in the unmanned aerial vehicle task unloading and resource scheduling process, and obtaining an optimal solution under the condition of meeting various constraint conditions.
2. The unmanned energy consumption optimization method based on the end-edge-cloud cooperative architecture and SHADE-NE algorithm according to claim 1, wherein the method comprises the following specific steps: Step S101, end-side-cloud architecture deployment and system initialization Firstly, completing architecture deployment of an end side, edge nodes and a cloud end in an unmanned aerial vehicle operation environment, and carrying out initialization configuration on a system, wherein a cloud control center is in communication connection with a plurality of edge servers, each edge server manages an edge management area; step S102, unmanned aerial vehicle management strategy based on end-side-cloud architecture In the step, the tasks generated by the unmanned aerial vehicle under the complex scene of inspection and disaster relief are managed and scheduled in a layered manner through an end-side-cloud three-layer cooperative mechanism, and the basic principle is that the unmanned aerial vehicle is used as an end side and has certain calculation and cache capability, so that real-time acquisition and partial calculation of data can be completed, an edge server has stronger calculation power and lower transmission delay and is suitable for processing tasks needing quick response, a cloud has global visual angle and strong centralized calculation capability and is suitable for processing tasks needing global optimization and large-scale data analysis, and the overall energy consumption can be reduced and the running stability of a system can be improved while the time delay constraint can be ensured by reasonably judging whether the tasks should be executed at the local and edge or at the cloud; In order to realize layered judgment of tasks, firstly, the communication conditions between the unmanned aerial vehicle and the edge and between the unmanned aerial vehicle and the cloud are considered, and the wireless transmission rate of the unmanned aerial vehicle and the edge server is expressed as follows: Wherein, the Is shown at the moment The transmission rate between the unmanned aerial vehicle and the edge server; channel bandwidth allocated between the unmanned aerial vehicle and the edge server; For the transmit power of the drone in the uplink, Is between unmanned plane and edge server at time Is used for the channel gain of (a), The noise power is the noise power of the receiving end; When the following criteria are satisfied: Wherein, the The minimum transmission rate threshold value is set for the system and is used for ensuring that task data is uploaded within an acceptable time delay range, wherein the threshold value is generally comprehensively determined by task time delay constraint, task data size and link reliability requirements; If the conditions are not satisfied, the task needs to be temporarily stored in the local area of the unmanned aerial vehicle, the cache module at the end of the unmanned aerial vehicle stores the task and performs partial operation, and the uploading is tried again when the subsequent link conditions are improved so as to avoid the task loss or overlarge delay; If the task runs locally on the unmanned aerial vehicle, the calculation time delay is as follows: The energy consumption is as follows: Wherein, the The computational delay created for the local operation of the drone, To offload the task proportion to the edge server, For the remaining proportion of tasks to be performed by the drone, In order to be able to carry out a task data volume, The number of CPU cycles required is calculated for the unit data, For the unmanned aerial vehicle CPU frequency, As a function of the energy consumption coefficient, The method comprises the steps of calculating energy consumption generated by locally executing tasks at time t for the unmanned aerial vehicle; when the link condition meets the unloading requirement, the task can be distributed to an edge server or cloud processing, and if the unmanned aerial vehicle is used for unloading the task to an edge end for operation, the calculation time delay of the unmanned aerial vehicle consists of transmission time delay and edge calculation time delay, and the calculation time delay is expressed as: Wherein, the The total computation delay incurred when offloading tasks to the edge servers for the drones, For the transmission delay of the unmanned aerial vehicle to the edge server transmission of the offloading task data, The computational delay incurred to process the offloaded tasks for the edge servers, To offload the task proportion to the edge server, In order to be able to carry out a task data volume, For the transmission rate between the drone and the edge server, The number of CPU cycles required to process the unit data for the edge server, CPU frequency of the edge server; the energy consumption that unmanned aerial vehicle produced in transmission process is: Wherein, the For the energy consumption generated by the unmanned aerial vehicle in the process of unloading the task to the edge server, For the transmit power of the unmanned aerial vehicle, To offload the task proportion to the edge server, In order to be able to carry out a task data volume, The transmission rate between the unmanned aerial vehicle and the edge server is set; if the task is further transmitted to the cloud, the total time delay is: The total time delay consists of three parts, namely the transmission time delay from the unmanned aerial vehicle to the edge, the transmission time delay from the edge to the cloud and the cloud computing time delay in sequence, The total computation delay generated when the unmanned aerial vehicle further offloads the task to the cloud end for execution, To offload the task proportion to the edge server, In order to be able to carry out a task data volume, For the transmission rate between the drone and the edge server, For the transmission rate between the edge server and the cloud, The number of CPU cycles required to process the unit data for the cloud, The CPU frequency of the cloud; The energy consumption is as follows: Wherein, the The unmanned aerial vehicle further offloads the task to the total energy consumption generated when the cloud is executed, For the transmit power of the unmanned aerial vehicle, To offload the task proportion to the edge server, In order to be able to carry out a task data volume, For the transmission rate between the drone and the edge server, The transmission rate between the edge server and the cloud end is set; finally, according to the criterion: Wherein, the The computational delay created for the local operation of the drone, Is the energy consumption of the unmanned aerial vehicle when in local execution, For the total delay in offloading tasks to the edge server for the drone, For communication energy consumption generated by the unmanned side when the task is offloaded to the edge server, To represent the total latency of a task when it is uploaded to the cloud, For the communication energy consumption generated by the unmanned aerial vehicle side when the task is uploaded to the cloud, The threshold is constrained for the time delay of the task, An energy consumption threshold set for the system; If only a single criterion condition is met, the task is executed according to the corresponding position, if a plurality of criterion conditions are met at the same time, the system introduces a comprehensive criterion for unified comparison and selection of different execution positions, wherein the comprehensive criterion is as follows: ; Wherein, the For the comprehensive evaluation index of the task at the kth execution position, Representing the total time delay of the task in the execution position, Representing the corresponding energy consumption of the device, And Respectively time delay constraint and energy consumption constraint of the task; Parameters and parameters The system is used for balancing the real-time performance and the energy efficiency of task execution and automatically improving the system under the condition of emphasizing the rapid completion of the task To ensure time delay performance, and under the condition of emphasizing endurance and energy efficiency, the system improves By the dynamic adjustment mechanism, the system can realize the self-adaptive optimization of the scheduling strategy under different tasks and environmental conditions, and the comprehensive requirements of time delay and energy efficiency are considered; Step S103, task scheduling and energy consumption optimization based on SHADE-NE algorithm In order to adapt to the environment of multiple unmanned aerial vehicles, global energy consumption optimization is realized under the complex scenes of the multiple unmanned aerial vehicles and the multiple tasks and the time delay requirement is met, a SHADE-NE algorithm is introduced in the step and used for carrying out joint optimization on task allocation and resource scheduling; First, the system is in time slot Is expressed as: Wherein, the Time slot for system Is used for the energy consumption of the (c) energy, For the task Energy consumption when executing locally at the drone, For the task The energy consumption when unloading to the edge end execution, For the task Unloading the energy consumption when the cloud is executed, wherein M is the total number of tasks to be processed in a time slot t of the system; in order to ensure that the task meets the constraint of time delay and energy consumption in the execution process, introducing an adaptability function with penalty term, which is expressed as: Wherein, the For a candidate solution Is used for the adaptation value of the (a), For the total energy consumption of the system, For the task Is provided with a delay in execution of (a), For the task Is used for the execution of the energy consumption of (a), For the maximum allowable time delay of a task, As a result of the energy consumption threshold value, Represents a penalty term when the constraint is violated, , The value range of the penalty coefficient can be set according to the task real-time requirement and the system energy consumption constraint condition; in the iterative updating process, the improved differential evolution algorithm adopts a current-to-pbest/1 mutation strategy to generate a new solution, and the formula is as follows Wherein, the A mutation vector generated by mutation operation for the ith individual in the g generation, Is the first Generation 1 The number of individuals who are to be treated, For the first few percent of the excellent individuals in the current population, , The individuals randomly selected from the current population or external archive respectively, Is a scaling factor; in the crossing stage, the variation vector and the current individual are recombined according to the crossing probability to obtain a test solution In the selection stage, the algorithm is used for adapting the degree function For evaluating indexes, comparing the test solution with the original individual, preferentially preserving, so as to gradually approach the global optimal solution, and after multiple generations of iteration, combining parameter self-adaptive adjustment, population scale reduction and restarting mechanisms, wherein the finally output optimal solution is as follows: Wherein, the For a system optimal scheduling solution obtained under the constraint condition of satisfying time delay and energy consumption, Represents the optimal execution position of the multi-task at the local, edge or cloud of the unmanned plane, Indicating an optimal power allocation scheme is shown, Indicating an optimal calculated frequency allocation scheme, By the scheme, the system can realize global energy consumption minimization in a multi-unmanned aerial vehicle multi-task environment on the premise of ensuring task time delay constraint.

Description

Unmanned energy consumption optimization method based on end-side-cloud cooperative architecture and SHADE-NE algorithm Technical Field The invention belongs to the technical field of unmanned aerial vehicle systems, and particularly relates to an unmanned aerial vehicle energy consumption optimization method based on an end-side-cloud cooperative architecture and SHADE-NE algorithm, which is applied to task unloading and resource scheduling of unmanned aerial vehicles in complex scenes such as power inspection, disaster relief, smart cities and the like. Background In recent years, with the wide application of unmanned aerial vehicles in the fields of inspection, disaster monitoring and the like, the data volume and the calculation demand generated in the task execution process are continuously increased. The traditional calculation mode depends on unmanned aerial vehicle local processing or single base station support, has the problems of poor real-time performance, high energy consumption, insufficient calculation power and the like, and is difficult to meet task requirements in complex environments. Therefore, a new architecture and an optimization mechanism are explored to improve the task execution efficiency of the unmanned aerial vehicle and reduce the whole energy consumption, and the method has important practical significance. In the prior art, in the invention patent CN202110135702.A, namely a method and a device for determining a task unloading strategy of an unmanned aerial vehicle based on reinforcement learning, the calculation task of the unmanned aerial vehicle is modeled as a directed acyclic graph, and the reinforcement learning method is utilized to determine whether a subtask is locally executed or unloaded to an edge server for processing, so that the time delay and the energy consumption of task execution are reduced to a certain extent. The scheme has certain advantages in terms of task decomposition and unloading decision. Meanwhile, reinforcement learning relies on a large number of training samples and environment interaction, the training cost is high, the instantaneity and the stability under a complex dynamic environment are difficult to ensure, and the overall scheduling efficiency and the energy consumption optimizing capability are limited. In the invention patent CN202311582845A 'a unmanned aerial vehicle task allocation and resource collaborative optimization method', by establishing an unmanned aerial vehicle task allocation model and combining constraint conditions of communication resources and computing resources, an optimization algorithm is adopted to carry out joint solution on task allocation and resource use, so that the unmanned aerial vehicle task execution efficiency is improved. The scheme provides certain improvements in resource allocation and task scheduling. Meanwhile, the optimization process is easy to sink into local optimum, the expansion capability of complex multitasking and multi-unmanned aerial vehicle cooperative scenes is insufficient, and the real-time performance and energy efficiency requirements in large-scale application are difficult to meet. Disclosure of Invention The invention aims to at least solve one of the technical defects, and provides an unmanned aerial vehicle energy consumption optimization method based on an end-side-cloud cooperative architecture and SHADE-NE algorithm, which can improve the real-time performance and reliability of unmanned aerial vehicle task processing, enhance the cooperative energy consumption optimization capability of multiple unmanned aerial vehicles and improve the stability and adaptability of the optimization process. In order to achieve the above purpose, the technical scheme adopted by the invention is that an unmanned energy consumption optimization method based on an end-side-cloud cooperative architecture and SHADE-NE algorithm is characterized by comprising the following steps: 1) And adopting an end-side-cloud cooperative framework, wherein a layered cooperative mechanism is established among the end side, the edge end and the cloud end of the unmanned aerial vehicle, the unmanned aerial vehicle is utilized to complete real-time acquisition and preliminary processing of data, and then the management strategy is utilized to judge the unloading position of the task, so that layered execution of task processing is realized. The method can effectively reduce the energy consumption and the transmission load of the unmanned aerial vehicle, improve the real-time performance and the reliability of task execution, and avoid the problems of delay and energy consumption caused by limited calculation force. 2) And adopting SHADE-NE optimization algorithm, solving in an iterative search mode in the unmanned aerial vehicle task unloading and resource scheduling process, and obtaining an optimal solution under the condition of meeting various constraint conditions. The method can dynamically judge the reaso