With the rapid growth of the number of unmanned aerial vehicles (UAVs), in order to assess the risk of people on the ground especially in cities, improve the operational efficiency of UAV, and reduce the threat of UAV to the crowd, it is necessary to make a specific path planning for UAVs based on the crowd density. However, static demographic data are still used as the ground risk analysis basis for the path planning of UAV, and real-time path planning of UAV according to the spatial-temporal characteristics of crowd density are often not performed. This study first analyzed the characteristics of urban road network crowd density based on the spatiotemporal data. Second, the convolutional neural network-long short-term memory (CNN-LSTM) combined model (C-Snet model) was established to predict the population density in different urban areas. Finally, the improved A* algorithm was used for real-time path planning and risk assessment of UAV according to the predicted crowd density data. The results show that the risk of UAV operation is reduced by 76%, which can provide a theoretical reference for the development of real-time path planning function of UAV traffic management system.
Unmanned aerial vehicle (UAV) is increasingly widely used, but with the continuous progress of urban development, the safe operation of UAV in cities is increasingly more prominent. Therefore, environmental risk assessment of obstacles has become one of the key issues in the field of low-altitude UAV research. In this study, taking the Beijing-Tianjin New Town as an example, the low-altitude airspace was divided into micro, light, and small UAV flight zones according to the low-altitude UAV types and operating heights. Based on the shape and size of UAVs, their motion constraints, and obstacle constraints, this research proposed an algorithm for approximation point expansion. The algorithm generates an expanded boundary on the basis of the original boundary of the obstacles, and this expanded boundary serves as a transitional zone between high risk and low risk areas in the low-altitude flight environment. Based on the UAV image data of 0.5 m resolution in the Beijing-Tianjin New Town in 2019, this study extracted obstacle elements in different assessment areas, and generated low-altitude flight obstacle environmental risk maps for different UAV types and different heights based on the risk assessment. The study area was divided into high-risk zone, high-risk transitional zone, medium-risk zone, and low-risk zone according to the threat posed to UAVs. The results show that: 1) The risk transitional zone in the micro, light, and small UAV control areas in the study area accounted for 10.9%, 7.3% and 9.0%, respectively, and the sharp-angle convex vertex optimization of the approximate point expansion algorithm can save about 1% of the airspace resources. 2) The proposed method can calculate the potential collision risk area of the UAVs in the flight area based on the mutual influence of the UAVs and the obstacles, and realize the effective assessment of the environmental risk of the low-altitude obstacles and provide a scientific reference for the navigability of the UAVs of different types in the flight area.
With the continuous opening of China's low-altitude airspace, relying on the existing low-altitude flight weather support technology to provide services for low-altitude safe flight is to some extent insufficient, and it is also difficult to forecast wind speed that has the greatest impact on flight. Based on the Weather Research and Forecasting (WRF) model mesoscale numerical model, this study simulated the wind speed and direction in the Beijing-Tianjin-Hebei region from 2015 to 2019, then compared the simulation results with the observation data of the weather stations in order to provide simulation tools for the safety of drones flying on low-altitude routes in the region. The main conclusions are as follows: the WRF model can better simulate the seasonal trend of wind speed, the simulation result in the plain areas is better than in the mountainous areas, and the simulated wind speed in the mountainous areas is higher than the observed data but the error is within an acceptable range (RMSE<1.5 m·s-1). The minimum values of average wind speed and maximum wind speed both appear in the late summer, and the maximum average wind speed appears in spring (4.43 m·s-1 in the mountainous areas, 4.13 m·s-1 in the plain areas). The maximum wind speed fluctuates and increases in winter, spring, and early summer, begins to decrease in mid-summer, and decreases to a minimum in late summer and early autumn. Wind speed in the Beijing-Tianjin-Hebei region is decreasing from northwest to southeast, the average wind speed at Potou Station (-0.02 m·s-1·(5 a)-1) and Tianjin Station (-0.02 m·s-1·(5 a)-1) showed a downward trend, the wind speed at other stations showed an upward trend, and Tangshan Station has the largest increase rate (0.08 m·s-1·(5 a)-1). With regard to the seasonal spatial distribution of wind speed, the average wind speed is mainly on the rise, and the station proportions are 45.45% in spring, 90.91% in summer, 63.63% in autumn, and 81.81% in winter. The prevailing wind in the plain areas is northeast-southwest; the wind direction of Huailai Station in the mountainous area is mainly in WNW direction (18.70%) and W direction (15.01%), while the wind direction of Yuxian Station is mainly in N (16.79%) and NNW (12.03%). Compared with the plain areas, the number of strong winds with wind speed of 8.0 m·s-1 has increased significantly in the mountainous areas. At a height of 1000 m, the frequency of strong winds in the plain areas increased significantly and the growth rate was higher than in the mountainous areas, which is not conducive to UAV flight, and the probability of occurrence of wind speeds above 17.0 m·s-1 is also significantly higher than in the mountainous areas.
In this study, the feasibility and effectiveness of unmanned aerial vehicle (UAV) detection system in civil airports were quantitatively evaluated according to the requirements of airport clearance management. First, multidimensional evaluation indicators were analyzed for UAV detection system, including the function, performance, reliability, safety, and deployability. Furthermore, ten system characteristics were considered for the evaluation, including monitoring coverage, effective detection range, spatial positioning accuracy, pitch accuracy, target tracking capability, real-time measurement, confidence degree of alerts, target recognition capability, reliability measure, information security, and deployability. By incorporating these ten characteristics, a comprehensive evaluation model was established based on fuzzy sets and fuzzy measures. The experimental results of multiple types of UAV detection system equipment were showed in real airport scenes. The results indicate that the proposed method can effectively measure the applicability and effectiveness of UAV detection system in civil airports, and has excellent compatibility with different types of technologies. The evaluation results can be well utilized for the selection of UAV detection system in civil airports.