340 / 2024-02-29 17:14:00

The Performance Evaluation of Flood Emergency Remote Sensing Monitoring Based on UAV swarm

performance Evaluation,remote sensing,UAV swarm,flood emergency monitoring

In recent years, the utilization of UAV for flood emergency remote sensing monitoring has increased. However, this field encounters various challenges, including harsh weather conditions, narrow time windows for emergency response, and the need for high data timeliness. One approach to tackle these challenges is through the use of UAV swarms, which involve multiple drones collaborating and making autonomous decisions to improve data collection efficiency. However, a significant challenge lies in evaluating the collaborative performance of UAV swarms in dynamic dam environments. Traditional evaluation methods, like the Analytic Hierarchy Process (AHP) that relies on expert experience, tend to decompose the dynamic collaboration process into multiple approximate and static processes. This approach disrupts the integrity and coherence of the swarm system in flood emergency remote sensing monitoring.

To overcome these limitations, this paper proposes a novel performance evaluation method that focuses on two crucial capabilities of UAV swarms: force allocation and route planning. The essential elements of this method are outlined as follows:



1.Construction of a flood prediction model using neural networks: A neural network model, specifically a convolutional neural network with Long Short-Term Memory(LSTM), is employed to analyze historical images linked to dam leakage-induced floods. This model extracts relevant information and enables real-time prediction of flood location and timing.



2.Design of a two-layer gridded simulation environment based on global aerial image: This simulation environment emphasizes dam topographic features and flood emergency time windows. These factors primarily influence the decision-making process regarding the number of UAV, the return period for aerial surveys, and the duration of UAV stays.



3.Development of a flood emergency monitoring performance evaluation model within the gridded simulation environment: This evaluation model dynamically assesses the performance of UAV swarms engaged in remote sensing monitoring activities, particularly in volatile and ever-changing environments. Two key indicators, namely task completion time and task completion degree, are employed to evaluate the efficiency and effectiveness of the UAVs in real-time.



By constructing a flood prediction model based on deep learning methods and employing dynamic performance evaluation, this method provides scientific support for real-time strategy optimization of UAV swarms. UAV swarms possess characteristics such as high sensitivity, robustness, and timeliness, making them a vital component in unmanned flood emergency monitoring. The proposed performance evaluation method will effectively promote the practical application of UAV swarm collaboration algorithms in flood emergency monitoring, leading to improved monitoring efficiency and facilitating broader utilization of UAV swarms in flood disaster prevention and reduction.