Unmanned aerial vehicle multi-height layer network coordination method and system based on low-altitude airspace operation risk

Abstract

The application discloses a UAV multi-height layer network cooperation method and system based on low-altitude airspace operation risk, and relates to the technical field of UAV cooperation control. The method comprises the following steps: collecting images in the low-altitude airspace, and dividing the images according to multiple height layers to obtain multiple height layer flying object densities; obtaining signal strengths controlled by UAVs in the multiple height layers, performing influence analysis and adjustment on the signal strengths based on the multiple height layer flying object densities to obtain multiple influence signal strengths; randomly setting multiple network parameter thresholds for UAV flight in the multiple height layers, and calculating to obtain a safety coefficient and a control coefficient; configuring a safety weight and a control weight according to the multiple height layer flying object densities and the multiple influence signal strengths, combining the safety coefficient and the control coefficient, calculating to obtain a cooperation fitness, and performing optimization processing to obtain multiple optimized network parameter thresholds to perform UAV cooperation control. The application effectively improves the safe and efficient cooperation control of a UAV group in a complex low-altitude environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle cooperative control, and particularly relates to a multi-altitude layer network cooperative method and system for unmanned aerial vehicles based on low-altitude airspace operation risk. BACKGROUND

[0002] With the wide popularity of unmanned aerial vehicles in logistics transportation, environmental monitoring, emergency rescue and other scenarios, the number of unmanned aerial vehicles in low-altitude airspace is increasing. When unmanned aerial vehicles operate in complex airspace, they are easily affected by flight object density, communication signal interference and other factors, thereby increasing the operation risk.

[0003] Existing unmanned aerial vehicle cooperative control methods mostly focus on optimization of a single altitude layer or a single-dimensional factor. It is difficult to cope with high-density and high-dynamic low-altitude environments using traditional methods, and the control cooperation is insufficient, resulting in poor unmanned aerial vehicle control effect and operation risk. SUMMARY

[0004] The present application provides a multi-altitude layer network cooperative method and system for unmanned aerial vehicles based on low-altitude airspace operation risk, aiming to solve the technical problems of insufficient unmanned aerial vehicle control cooperation and poor control effect in the prior art.

[0005] In view of the above problems, the present application provides a multi-altitude layer network cooperative method and system for unmanned aerial vehicles based on low-altitude airspace operation risk.

[0006] In a first aspect, the present application provides a multi-altitude layer network cooperative method for unmanned aerial vehicles based on low-altitude airspace operation risk, comprising:

[0007] Collecting images in low-altitude airspace and dividing them according to multiple altitude layers, identifying flight object density, and obtaining multiple altitude layer flight object densities;

[0008] Obtaining multiple signal strengths controlled by unmanned aerial vehicles at multiple altitude layers, performing influence analysis and adjustment on the multiple signal strengths based on the multiple altitude layer flight object densities, and obtaining multiple influence signal strengths;

[0009] Randomly setting multiple network parameter thresholds for unmanned aerial vehicle flight in multiple altitude layers, and calculating multiple safety coefficients and multiple control coefficients;

[0010] According to the multiple altitude layer flight object densities and the multiple influence signal strengths, configuring multiple safety weights and control weights, combining the multiple safety coefficients and the multiple control coefficients, calculating a cooperative fitness, and optimizing the multiple optimized network parameter thresholds to perform unmanned aerial vehicle cooperative control.

[0011] In a second aspect, the present application provides a multi-altitude layer network cooperative system for unmanned aerial vehicles based on low-altitude airspace operation risk, comprising:

[0012] The low-altitude image layering module is configured to collect images in the low-altitude airspace, divide the images according to a plurality of height layers, identify the density of flying objects in the plurality of height layers, and obtain the density of flying objects in the plurality of height layers.

[0013] The signal strength influence analysis module is configured to obtain a plurality of signal strengths controlled by the unmanned aerial vehicles in the plurality of height layers, perform influence analysis and adjustment on the plurality of signal strengths based on the density of flying objects in the plurality of height layers, and obtain a plurality of influence signal strengths.

[0014] The network parameter evaluation module is configured to randomly set a plurality of network parameter thresholds for the unmanned aerial vehicles flying in the plurality of height layers, and calculate a plurality of safety coefficients and a plurality of control coefficients.

[0015] The cooperative optimization decision module is configured to configure a plurality of safety weights and a plurality of control weights according to the density of flying objects in the plurality of height layers and the plurality of influence signal strengths, combine the plurality of safety coefficients and the plurality of control coefficients, calculate a cooperative fitness, and perform optimization processing to obtain a plurality of optimized network parameter thresholds for cooperative control of the unmanned aerial vehicles.

[0016] One or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0017] The present application provides a method and system for unmanned aerial vehicle multi-height layer network cooperation based on low-altitude airspace operation risk, which can realize intelligent management and control of unmanned aerial vehicle groups in complex airspace environments. The present application realizes a method for unmanned aerial vehicle cooperative control that can adapt to complex low-altitude environments and intelligently balance safety and efficiency by organically integrating airspace perception, communication analysis, parameter optimization, and cooperative control, effectively improving unmanned aerial vehicle control effect and improving the safety of unmanned aerial vehicle operation in low-altitude airspace. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0019] Figure 1 The flowchart of the method for unmanned aerial vehicle multi-height layer network cooperation based on low-altitude airspace operation risk provided by the embodiment of the present application is shown.

[0020] Figure 2 The structure diagram of the system for unmanned aerial vehicle multi-height layer network cooperation based on low-altitude airspace operation risk provided by the embodiment of the present application is shown.

[0021] In the drawings, the components represented by the numbers are described as follows:

[0022] The low-altitude image layering module 11, the signal strength influence analysis module 12, the network parameter evaluation module 13, and the collaborative optimization decision module 14. DETAILED DESCRIPTION

[0023] The application provides a UAV multi-height layer network collaboration method based on low-altitude airspace operation risk, which is used to solve the technical problems of insufficient UAV control collaboration and poor control effect in the prior art.

[0024] The technical solutions in the embodiments of the application will be clearly and completely described in connection with the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the application.

[0025] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion, for example, a process, method, system, product or server comprising a series of steps or units need not be limited to only those steps or units clearly listed, but can include other steps or modules not clearly listed or inherent to the process, method, product or device.

[0026] Embodiment one, as shown in the application provides a UAV multi-height layer network collaboration method based on low-altitude airspace operation risk, which comprises: Figure 1

[0027] S100: Collect images in the low-altitude airspace, and divide according to multiple height layers to identify the density of flying objects and obtain the density of flying objects in multiple height layers.

[0028] In the embodiments of the application, images in the low-altitude airspace are collected, and divided according to multiple height layers to identify the density of flying objects and obtain the density of flying objects in multiple height layers. Through visual sensing technology, the detection of low-altitude airspace flying targets and the quantitative evaluation of different height layer risk levels are realized; the deep learning model is used to automatically identify the density, avoiding tedious and inefficient manual observation; the intuitive image is converted into quantifiable density values, providing accurate data basis for subsequent risk assessment and decision-making.

[0029] Specifically, the step S100 comprises the following sub-steps:

[0030] Collect images in the low-altitude airspace to obtain low-altitude images;

[0031] Divide the low-altitude images according to multiple height layers to obtain multiple height layer images;

[0032] ​The multiple height layer images are respectively input into the flying object density identifier, and multiple height layer flying object densities are obtained through identification output.

[0033] In the embodiments of the present application, first, images in the low-altitude airspace are collected to obtain low-altitude images. The images can provide key information such as spatial distribution, motion trajectory, target size and type, and are the basis for subsequent division by height layer and density estimation; through high-quality collection to reduce detection errors, the accuracy and robustness of subsequent identification and statistics can be improved. For example, image collection devices such as ground fixed cameras and unmanned aerial cameras are deployed in the target low-altitude airspace, and low-altitude images covering the target low-altitude airspace are continuously or periodically collected. High-resolution, wide-dynamic-range cameras can be used to cope with complex lighting and weather conditions, thereby obtaining high-precision low-altitude images to ensure the accuracy of subsequent analysis.

[0034] Then, the low-altitude images are divided according to multiple height layers to obtain multiple height layer images. The risk, target type and communication influence of the low-altitude airspace often differ significantly with height, for example, 0-50m is often for birds or low-flying drones, and 50-100m is for aerial photography and delivery drone activity layers. In the park low-altitude monitoring scene, first, a signal tower in the park is selected, and the tower tip elevation is measured with an RTK positioning instrument. RTK (Real-time kinematic) is a measurement method that can obtain centimeter-level positioning accuracy in the field in real time. The signal tower with an altitude of 150m is known as a calibration reference, and the tower tip pixel coordinates (350, 420) are marked in the low-altitude image. Then, the camera calibration parameters are used: focal length 1000 pixels, principal point coordinates 600, 400, and installation angle 10° downward, combined with the perspective principle to establish a pixel-elevation mapping. For example, the pixel coordinates (280, 310) in the image correspond to an altitude of 72m through mapping calculation, which is classified into the 50-100m height layer. Finally, all pixels are classified into corresponding layers according to the calculated calibration elevation at intervals of 0-50m, 50-100m, …, 250-300m to generate 6 height layer images, such as the 0-50m layer having tree tops and low-flying bird pixels, and the 50-100m layer having delivery drone pixels, facilitating targeted control.

[0035] Finally, the multiple height layer images are respectively input into the flying object density identifier, and multiple height layer flying object densities are obtained through identification output. The flying object density in each height layer is a key indicator for quantifying airspace congestion and interference risk. Density is more robust than simple target detection, and when targets overlap or have large size differences, density estimation is often more reliable than individual detection. The output of the flying object density values or density maps of each height layer can provide direct quantitative input for subsequent signal strength influence analysis, network parameter evaluation, and collaborative optimization decision calculation.

[0036] The training step of the flying object density identifier includes:

[0037] First, a sample height layer image set is collected, the flying object density in each sample height layer image is labeled, and a sample flying object density set is obtained. A large number of sample height layer image sets containing different scenes, different weathers, and different flying objects are collected, the flying objects in each image are labeled by professional personnel, point labeling or frame labeling can be used, and the real sample flying object density set is calculated according to the number of labeled points and the effective area of the image. Among them, point annotation refers to labeling a point at the center of each flying object, which is suitable for density estimation; frame labeling refers to drawing a frame for each target, which is suitable for detector training.

[0038] Then, a flying object density identifier is constructed based on a convolutional neural network (CNN). The convolutional neural network can use an encoder-decoder structure for density map estimation, or a mainstream target detection network to detect targets and then calculate the density, such as YOLO, Faster R-CNN, or directly use an end-to-end density regression network, such as CSRNet.

[0039] Finally, the sample height layer image set and the sample flying object density set are used to supervise the training of the flying object density identifier until convergence. The sample height layer image set is used as input, the sample flying object density set is used as target label, the mean square error (MSE) or the mean absolute error (MAE) is used as the loss function, the constructed neural network is supervised and trained, the network weights are continuously optimized through the back propagation algorithm, and the precision of the model on the validation set converges to meet the application requirements.

[0040] The training of the flying object density identifier enables the model to learn effective features from specific scenes, thereby achieving high-precision density estimation or target detection in actual deployment. The CNN model can effectively learn the abstract features of flying objects, has good recognition ability for small and fuzzy targets, and has strong anti-interference ability. Through supervised learning, a robust model in the scene is obtained, the false positive or false negative rate is reduced, and the credibility is improved.

[0041] S200: Obtain a plurality of signal strengths of a plurality of unmanned aerial vehicles controlled at a plurality of height layers, and perform influence analysis and adjustment on the plurality of signal strengths based on the flying object density at the plurality of height layers to obtain a plurality of influence signal strengths.

[0042] In the embodiment of the present application, the signal strengths controlled by the unmanned aerial vehicle at multiple height layers are obtained, the signal strengths are adjusted based on the influence analysis of the flying object density at multiple height layers, and multiple influence signal strengths are obtained. When the unmanned aerial vehicle flies at low altitude, the control link is easily affected by the airspace environment. When the flying object density is large, problems such as decrease in received power, fluctuation in received signal, increase in packet loss rate, and increase in delay may occur. This step corrects the original signal strength based on the flying object density, analyzes the amplitude of the influence signal strength, adjusts the signal strength, and obtains multiple influence signal strengths to more truly reflect the communication quality at multiple height layers.

[0043] Specifically, the step S200 includes the following sub-steps:

[0044] The average of the signal strengths controlled by the unmanned aerial vehicle at multiple height layers is tested and obtained as multiple signal strengths.

[0045] The flying object density at each height layer is input into a signal influence classification table, and multiple signal influence coefficients are obtained.

[0046] The multiple signal strengths are adjusted by using the multiple signal influence coefficients, and multiple influence signal strengths are obtained.

[0047] First, the average of the signal strengths controlled by the unmanned aerial vehicle at multiple height layers is tested and obtained as multiple signal strengths. The unmanned aerial vehicle or test node is deployed at different height layers to test the signal strength of the communication link between the unmanned aerial vehicle and the control station at multiple height layers. The signal strength is a general concept, which can include but is not limited to the following quantifiable technical indicators: received signal strength indication (RSSI), signal-to-noise ratio (SNR), packet loss rate, etc. The average of the test values of multiple sampling points in each height layer is obtained to obtain the signal strength average corresponding to each height layer as the initial signal strength value of the layer. For example, the average RSSI is -65dBm and the packet loss rate is 3% at a height of 50m, and the average RSSI is -72dBm and the packet loss rate is 7% at a height of 100m. The higher the value of the received signal strength indication RSSI, the stronger the signal; the packet loss rate is the proportion of lost data packets in the transmission process, and the higher the packet loss rate, the worse the communication quality. This step converts the continuous signal fluctuation into stable data representing the overall situation of each layer.

[0048] Then, each height layer flying object density input signal influence classification table, output a plurality of signal impact coefficient. Signal impact classification table can show the mapping relationship between different height layer flying object density range and signal impact coefficient. The classification table of flying object density-signal impact coefficient is established in advance, which can be obtained by experimental test or simulation. Each height layer flying object density input signal influence classification table, output a plurality of signal impact coefficient. Signal impact coefficient is a dimensionless scaling factor, which is used to represent the amplitude of signal strength attenuation or fluctuation due to shielding and interference, the smaller the signal impact coefficient, the greater the impact. If the signal impact coefficient < 1.0, it means signal attenuation; if the signal impact coefficient > 1.0, it means there may be signal enhancement; usually signal impact coefficient ≤ 1.0. For example, in the 0-50m layer, the flying object density is 0.6 / 100㎡, which is input into the signal impact classification table, and the corresponding signal impact coefficient is 0.70; in the 50-100m layer, the flying object density is 0.2 / 100㎡, which is input into the signal impact classification table, and the corresponding signal impact coefficient is 0.85.

[0049] Finally, a plurality of signal impact coefficients are used to calculate and adjust a plurality of signal strengths to obtain a plurality of impact signal strengths. The calculation formula can be simplified as: impact signal strength = RSSI + (1-signal impact coefficient) x RSSI. Through the above calculation, a plurality of impact signal strengths can be obtained. The value is a theoretical value after environmental factor correction, which can better reflect the actual available communication quality of the unmanned aerial vehicle in the height layer, and provides a basis for subsequent cooperative control strategy. For example, if in the 0-50m layer, the original RSSI is-65dBm, the packet loss rate is 3%, the flying object density is 0.6, and the signal impact coefficient is 0.7, then the corrected impact signal strength is-65+(1-0.70)x(-65)=-84.5dBm; in the 50-100m height layer, the original RSSI is-72dBm, the packet loss rate is 7%, the flying object density is 0.2, and the signal impact coefficient is 0.85, then the corrected signal strength is-72+(1-0.85)x(-72)=-82.8dBm. As can be seen, although the original RSSI in the 50-100m layer is lower, the corrected signal quality is still better than that in the 0-50m layer, so the 100m layer flight is preferred.

[0050] In the embodiments of the present application, a method for predicting signal quality is provided. Compared with directly measuring signal strength, the influence signal strength corrected by density can better reflect the risk situation. If only the original signal strength is considered, the risk of communication deterioration caused by the increase of flying objects can be ignored. The influence signal strength is taken as an input, which provides more accurate communication state data input after normalization and environmental compensation for subsequent collaborative optimization decision, and can improve the safety and stability of the unmanned aerial vehicle cluster. Before the unmanned aerial vehicle enters a certain airspace with high density, the possible interference on communication can be predicted in advance, and active measures can be taken, rather than passive reaction after communication deterioration.

[0051] S300: Randomly set a plurality of network parameter thresholds for the unmanned aerial vehicles flying in a plurality of height layers, and calculate a plurality of safety coefficients and a plurality of control coefficients.

[0052] In the embodiments of the present application, a plurality of network parameter thresholds for the unmanned aerial vehicles flying in a plurality of height layers are randomly set, and a plurality of safety coefficients and a plurality of control coefficients are calculated. The network parameter threshold, such as a signal-to-noise ratio threshold, can control the unmanned aerial vehicle when the control signal meets the threshold, so as to ensure the control quality, otherwise the unmanned aerial vehicle will hover or automatically avoid obstacles and return to the starting point. By randomly initializing a series of network parameter combinations and based on a novel evaluation strategy based on historical experience, the potential performance of each parameter combination in safety and controllability is quickly calculated, which provides a quantitative basis for subsequent optimization decision.

[0053] Specifically, the step S300 includes the following sub-steps:

[0054] The network parameter thresholds for the unmanned aerial vehicles flying in a plurality of height layers are randomly set, wherein the network parameter thresholds include signal-to-noise ratio thresholds;

[0055] An average network parameter threshold set in the past time is obtained;

[0056] The ratios of the plurality of network parameter thresholds to the average network parameter threshold are calculated respectively, and a plurality of safety coefficients are obtained;

[0057] The ratios of the average network parameter threshold to the plurality of network parameter thresholds are calculated respectively, and a plurality of control coefficients are obtained.

[0058] First, a plurality of network parameter thresholds for the plurality of altitude layers are randomly set, wherein the network parameter thresholds include a signal-to-noise ratio threshold. Random setting is to sample in a reasonable parameter space to explore a plurality of possible configuration schemes to obtain optimal unmanned aerial vehicle control network parameters, which is the principle in the optimization algorithm. The signal-to-noise ratio (SNR) is the ratio of signal power to noise power, usually expressed in dB, and the higher the signal-to-noise ratio, the better the link quality. The signal-to-noise ratio threshold determines the minimum signal quality threshold required for the unmanned aerial vehicle to reliably receive and execute control instructions. Each random setting will generate a set of network parameter threshold configurations covering all altitude layers, and at each altitude layer, a threshold is randomly generated as the minimum control link requirement for the unmanned aerial vehicle. For example, the random signal-to-noise ratio threshold is 18 dB at 0-50 m, 15 dB at 50-100 m, and 20 dB at 100-150 m.

[0059] Then, the average network parameter threshold set in the past time is obtained. From the historical configuration record, the average network parameter threshold successfully applied in the past period of time and not causing an accident is obtained, which is used as a benchmark value representing the historical average safety level, which can reflect the relatively conservative and stable configuration adopted by the unmanned aerial vehicle in most cases in the past. For example, the historical average signal-to-noise ratio threshold is 16 dB. The historical average signal-to-noise ratio threshold is used as a comparison benchmark, which can reflect the global control standard and avoid extreme deviation caused by a single random value.

[0060] Secondly, the ratio of the plurality of network parameter thresholds to the average network parameter threshold is calculated to obtain a plurality of safety coefficients. Safety coefficient = random set network parameter threshold / average network parameter threshold, the ratio is used to quantify the relative strictness of the altitude layer in safety. For example, the average signal-to-noise ratio threshold is 16 dB; in the 0-50 m layer, the random signal-to-noise ratio threshold is 18 dB, the safety coefficient = 18 ÷ 16 = 1.125; in the 50-100 m layer, the random signal-to-noise ratio threshold is 15 dB, the safety coefficient = 15 ÷ 16 = 0.937; in the 100-150 m layer, the random signal-to-noise ratio threshold is 20 dB, the safety coefficient = 20 ÷ 16 = 1.25. The safety coefficient > 1 indicates that the layer has a higher signal quality requirement for the unmanned aerial vehicle and pays more attention to safety; the safety coefficient < 1 indicates that the layer has a lower safety requirement and the unmanned aerial vehicle has a risk of losing control.

[0061] Finally, the ratio of the average network parameter threshold and the plurality of network parameter thresholds is calculated respectively to obtain a plurality of control coefficients. The control coefficient = average network parameter threshold / randomly set network parameter threshold. The ratio is another expression of the safety coefficient inverse, which is used to quantify the relative ease of use of the altitude layer in task control flexibility. The greater the current set signal-to-noise ratio threshold, the lower the control coefficient. For example, in the 0-50m layer, the control coefficient = 16 ÷ 18 ≈ 0.89; in the 50-100m layer: control coefficient = 16 ÷ 15 ≈ 1.07; in the 100-150m layer, the control coefficient = 16 ÷ 20 = 0.80. If the control coefficient < 1, it means that the control is more conservative, and the unmanned aerial vehicle is limited by the signal strength; if the control coefficient > 1, it means that the control is more flexible, and allows the unmanned aerial vehicle to work in a worse signal.

[0062] In the embodiments of the present application, the randomization threshold can cover different environmental fluctuation scenarios, avoiding overfitting under a single parameter; using historical safety records as a benchmark makes the evaluation results have actual basis, ensuring the rationality and reliability of the evaluation results; the safety coefficient can reflect the degree of overestimation of the current threshold relative to the average level, and a high safety coefficient indicates a more stringent safety requirement; the control coefficient can reflect the control flexibility under the condition of meeting the average threshold, and a high control coefficient indicates that the task is easier to execute; the safety coefficient and the control coefficient can be used for weighting in subsequent steps to improve the adaptability of cooperative control.

[0063] S400: According to the plurality of altitude layer flying object densities and the plurality of influence signal strengths, a plurality of safety weights and control weights are configured, a plurality of safety coefficients and a plurality of control coefficients are combined, a cooperative adaptability is calculated and obtained, and a plurality of optimized network parameter thresholds are obtained by optimization processing, for unmanned aerial vehicle cooperative control.

[0064] In the embodiments of the present application, the higher the flying object density, the greater the collision risk, and the safety priority should be improved; the stronger the signal strength of the control link, the better the communication quality of the altitude layer, and the control reliability weight should be increased accordingly; through the distribution of safety weight and control weight, the unmanned aerial vehicle cooperative control realizes dynamic balance between safety and efficiency; the cooperative adaptability calculation can be used as an evaluation index to guide the iterative optimization of the network parameter threshold, and finally the optimal flight configuration is obtained.

[0065] Specifically, the present step S400 includes the following sub-steps:

[0066] The ratio of the plurality of flying object densities and the preset flying object density is calculated as a plurality of density coefficients;

[0067] The ratio of the preset signal strength and the plurality of influence signal strengths is calculated as a plurality of signal coefficients;

[0068] According to the plurality of density coefficients and the plurality of signal coefficients, the plurality of safety weights and the plurality of control weights are obtained by distribution calculation.

[0069] According to the plurality of safety weights and the plurality of control weights, the plurality of safety coefficients and the plurality of control coefficients are obtained by weighted calculation.

[0070] First, the ratio of the plurality of flying object densities and the preset flying object density is calculated as the plurality of density coefficients; for each height layer, the ratio of the flying object density of the layer and a preset flying object density is calculated as the density coefficient of the layer. Density coefficient = flying object density of the height layer / preset flying object density. The density coefficient > 1 indicates that the risk of the layer is higher than the preset level, and the greater the density coefficient, the higher the risk. For example, in the 0-50m layer, the flying object density is 0.8 / 100m 2 , and the preset flying object density is 0.5 / 100m 2 , then the density coefficient = 0.8 / 0.5 = 1.6; in the 50-100m layer, the flying object density is 0.3 / 100m 2 , and the preset flying object density is 0.5 / 100m 2 , then the density coefficient = 0.3 / 0.5 = 0.6.

[0071] The ratio of the preset signal strength and the plurality of influence signal strengths is calculated as the plurality of signal coefficients. For each height layer, the ratio of a preset signal strength and the influence signal strength of the layer is calculated as the signal coefficient of the layer. Signal coefficient = preset signal strength / influence signal strength. The signal coefficient > 1 indicates that the actual signal quality is lower than the preset standard, the communication reliability is poor, and the greater the signal coefficient, the worse the communication environment. For example, in the 0-50m layer, the influence signal strength is -45.5dBm, the preset signal strength is -50dBm, and the signal coefficient is (-50) / (-45.5)≈1.10; in the 50-100m layer, the influence signal strength is -61.2dBm, the preset signal strength is -50dBm, and the signal coefficient is (-50) / (-61.2)≈0.82.

[0072] Then, according to the plurality of density coefficients and the plurality of signal coefficients, a plurality of safety weights and a plurality of control weights are calculated and obtained by distribution. The normalization method is used to independently calculate the weight of each height layer. Safety weight = density coefficient / (density coefficient + signal coefficient); control weight = signal coefficient / (density coefficient + signal coefficient). For example, in the 0-50m layer, the density coefficient is 1.6, and the signal coefficient is 1.10, so the safety weight = 1.6 / (1.6 + 1.10) = 0.59, and the control weight = 1.10 / (1.6 + 1.10) = 0.41; in the 50-100m layer, the density coefficient is 0.6, and the signal coefficient is 0.82, so the safety weight = 0.6 / (0.6 + 0.82) = 0.42, and the control weight = 0.82 / (0.6 + 0.82) = 0.58. The greater the density of the flying object, the greater the density coefficient, and the greater the safety weight, which ensures that the safety decision has a higher priority in the high-risk airspace. The smaller the influence signal strength, the greater the signal coefficient, and the greater the control weight, indicating that in the airspace with poor communication conditions, more attention should be paid to the stability of control rather than pursuing the control efficiency; when the signal is very good, the control weight is smaller, allowing pursuit of control efficiency.

[0073] According to the plurality of safety weights and the plurality of control weights, a plurality of safety coefficients and a plurality of control coefficients are weighted and calculated to obtain a cooperative fitness. The formula is: cooperative fitness = safety weight × safety coefficient + control weight × control coefficient. For example, in the 0-50m layer, the safety weight is 0.59, the safety coefficient is 1.125, the control weight is 0.41, and the control coefficient is 0.89, so the cooperative fitness = 0.59 × 1.125 + 0.41 × 0.89 ≈ 1.03; in the 50-100m layer, the safety weight is 0.42, the safety coefficient is 0.937, the control weight is 0.58, and the control coefficient is 1.07, so the cooperative fitness = 0.42 × 0.937 + 0.58 × 1.07 ≈ 1.01. The cooperative fitness value comprehensively evaluates the comprehensive performance of a set of network parameter threshold configurations under the current airspace environment and communication conditions. The higher the cooperative fitness value, the better the balance between safety and control of the set of parameters.

[0074] Finally, a plurality of optimized network parameter thresholds are obtained by optimization processing for unmanned aerial vehicle cooperative control, including:

[0075] Continue to randomly set multiple network parameter thresholds for the flight of the unmanned aerial vehicle in multiple height layers, perform iterative optimization, and obtain the multiple network parameter thresholds corresponding to the maximum synergy fitness after convergence as the multiple optimized network parameter thresholds for the cooperative control of the unmanned aerial vehicles. Randomly initialize the network parameter thresholds of the multiple height layers, such as the signal-to-noise ratio thresholds; perform iterative updating, calculate the corresponding safety coefficients, control coefficients, safety weights, control weights, and synergy fitness; use optimization algorithms such as genetic algorithms, particle swarm optimization, or gradient search to constantly find the parameter combination that maximizes the overall synergy fitness; and obtain the network parameter thresholds corresponding to the maximum synergy fitness after convergence as the final optimization result. For example, the initial signal-to-noise ratio thresholds of a height layer are {18, 15, 20} dB, and after 30 iterations of optimization, the signal-to-noise ratio thresholds corresponding to the maximum synergy fitness point become {17, 16, 19} dB. Applying this combination to the unmanned aerial vehicle group can achieve cooperative control of multiple height layers.

[0076] In the embodiments of the present application, the flight object density, signal strength, safety coefficient, and control coefficient are unified into the same framework to comprehensively reflect the risks and performance; the safety weight and control weight change in real time with the environmental state, having true environmental perception and adaptive ability; through iterative optimization, the network parameter thresholds of the unmanned aerial vehicles converge to the maximum synergy fitness point, improving the overall operating efficiency and safety; and the final output is a complete set of optimized parameters for all height layers in the entire airspace, realizing the leap from single-machine control to integrated cooperative control.

[0077] Embodiment two, as shown in Figure 2 The present application provides a multi-height layer network cooperative system for unmanned aerial vehicles based on low-altitude airspace operation risks, which comprises:

[0078] A low-altitude image layering module 11 is configured to collect images in the low-altitude airspace and divide them according to multiple height layers, identify the flight object density, and obtain the flight object density of multiple height layers;

[0079] A signal strength influence analysis module 12 is configured to obtain multiple signal strengths controlled by the unmanned aerial vehicles at multiple height layers, perform influence analysis and adjustment on the multiple signal strengths based on the flight object density of multiple height layers, and obtain multiple influence signal strengths;

[0080] A network parameter evaluation module 13 is configured to randomly set multiple network parameter thresholds for the flight of the unmanned aerial vehicle in multiple height layers, and calculate the multiple safety coefficients and multiple control coefficients;

[0081] A cooperative optimization decision module 14 is configured to configure multiple safety weights and control weights according to the flight object density of multiple height layers and the multiple influence signal strengths, combine the multiple safety coefficients and multiple control coefficients, calculate the synergy fitness, and perform optimization processing to obtain multiple optimized network parameter thresholds for the cooperative control of the unmanned aerial vehicles.

[0082] In an embodiment, the low-altitude image layering module 11 is further configured to:

[0083] collect images within the low-altitude airspace to obtain low-altitude images;

[0084] divide the low-altitude images according to a plurality of height layers to obtain a plurality of height layer images;

[0085] input the plurality of height layer images into a flying object density identifier respectively to obtain a plurality of height layer flying object densities by identification output;

[0086] wherein the training step of the flying object density identifier comprises:

[0087] collect a sample height layer image set, and label the flying object density in each sample height layer image to obtain a sample flying object density set;

[0088] construct a flying object density identifier based on a convolutional neural network;

[0089] use the sample height layer image set and the sample flying object density set to supervise training of the flying object density identifier until convergence.

[0090] In an embodiment, the signal strength influence analysis module 12 is further configured to:

[0091] test the average signal strength of the unmanned aerial vehicle control under a plurality of height layers as a plurality of signal strengths;

[0092] input the flying object density of each height layer into a signal influence classification table to obtain a plurality of signal influence coefficients by output;

[0093] use the plurality of signal influence coefficients to perform influence calculation and adjustment on the plurality of signal strengths to obtain a plurality of influence signal strengths.

[0094] In an embodiment, the network parameter evaluation module 13 is further configured to:

[0095] randomly set a plurality of network parameter thresholds for the unmanned aerial vehicle flight in the plurality of height layers, wherein the network parameter thresholds include a signal-to-noise ratio threshold;

[0096] obtain the average network parameter threshold set in the past time;

[0097] calculate the ratio of each network parameter threshold to the average network parameter threshold to obtain a plurality of safety coefficients;

[0098] calculate the ratio of the average network parameter threshold to the plurality of network parameter thresholds to obtain a plurality of control coefficients.

[0099] In one embodiment, the cooperative optimization decision module 14 is further configured to:

[0100] calculate a ratio of the plurality of flying object densities and the preset flying object density as a plurality of density coefficients;

[0101] calculate a ratio of the preset signal intensity and the plurality of influence signal intensities as a plurality of signal coefficients;

[0102] distribute the plurality of density coefficients and the plurality of signal coefficients to obtain a plurality of safety weights and a plurality of control weights;

[0103] weight the plurality of safety coefficients and the plurality of control coefficients according to the plurality of safety weights and the plurality of control weights to obtain a cooperative fitness.

[0104] wherein the optimization processing obtains a plurality of optimized network parameter thresholds for cooperative control of the UAVs, including:

[0105] continuously randomly setting the plurality of network parameter thresholds for the UAVs to fly in the plurality of height layers, iteratively optimizing, and after convergence, obtaining the plurality of network parameter thresholds corresponding to the maximum cooperative fitness as the plurality of optimized network parameter thresholds for cooperative control of the UAVs.

[0106] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. And the above describes the specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or can be advantageous.

[0107] The above only describes the preferred embodiments of the present application, and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0108] The present application and the drawings are only exemplary descriptions of the present application, and are considered to cover any and all modifications, changes, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art can make various modifications and changes to the present application without departing from the scope of the present application. Thus, if these modifications and changes of the present application belong to the scope of the present application and its equivalents, the present application intends to include these modifications and changes.

Claims

1. A multi-altitude layer network cooperative method for unmanned aerial vehicles (UAVs) based on low-altitude airspace operational risks, characterized in that, The method includes: Images of the low-altitude airspace are acquired and divided into multiple altitude layers to identify the density of flying objects and obtain the density of flying objects at multiple altitude layers. The system acquires multiple signal strengths for UAV control at multiple altitude levels, performs impact analysis and adjustment on these signal strengths based on the density of flying objects at multiple altitude levels, and obtains multiple influencing signal strengths. Multiple network parameter thresholds are randomly set for UAV flight at multiple altitude levels, and multiple safety factors and multiple control factors are calculated, including: Multiple network parameter thresholds are randomly set for UAV flight at multiple altitude levels, including a signal-to-noise ratio threshold. Obtain the average network parameter threshold set over a past period; Calculate the ratio of multiple network parameter thresholds to the average network parameter threshold to obtain multiple security factors; Calculate the ratio of the average network parameter threshold and multiple network parameter thresholds to obtain multiple manipulation coefficients; Based on the density of flying objects at multiple altitude levels and the strength of multiple influencing signals, multiple safety weights and control weights are configured. Combined with multiple safety coefficients and multiple control coefficients, a cooperative fitness degree is calculated, and multiple optimized network parameter thresholds are obtained through optimization processing. This is then used for UAV cooperative control, including: Calculate the ratio of multiple object densities to a preset object density, and use these ratios as multiple density coefficients; Calculate the ratio of the preset signal strength to multiple influencing signal strengths, and use these ratios as multiple signal coefficients; Based on multiple density coefficients and multiple signal coefficients, multiple security weights and multiple control weights are calculated and assigned. The cooperative fitness degree is obtained by weighting multiple safety coefficients and multiple control coefficients according to the multiple safety weights and multiple control weights.

2. The UAV multi-altitude layer network cooperative method based on low-altitude airspace operation risk according to claim 1, characterized in that, Images of the low-altitude airspace are acquired and divided into multiple altitude layers. Airborne object density is identified and obtained at these multiple altitude layers, including: Collect images within the low-altitude airspace to obtain low-altitude images; The low-altitude image is divided into multiple altitude layers to obtain multiple altitude layer images; Multiple altitude layer images are input into the flying object density recognizer, and the recognition output obtains the flying object density at multiple altitude layers.

3. The UAV multi-altitude layer network cooperation method based on low-altitude airspace operation risk according to claim 2, characterized in that, The training steps for the flying object density identifier include: Collect a set of sample altitude layer images, label the density of flying objects in each sample altitude layer image, and obtain a set of sample flying object densities; Construct a flight object density recognizer based on convolutional neural networks; The object density recognizer is trained under supervised supervision using the sample height layer image set and the sample object density set until convergence.

4. The UAV multi-altitude layer network cooperative method based on low-altitude airspace operation risk according to claim 1, characterized in that, Multiple signal strengths for UAV control at various altitude levels are acquired. Based on the density of flying objects at multiple altitude levels, the influence of these signal strengths is analyzed and adjusted to obtain multiple influencing signal strengths, including: The test obtained the average signal strength of the drone control at multiple altitude levels, which were used as multiple signal strengths. Input the density of flying objects at each altitude level into the signal influence classification table, and output multiple signal influence coefficients. Multiple signal influence coefficients are used to calculate and adjust the influence of multiple signal strengths, thereby obtaining multiple influencing signal strengths.

5. The UAV multi-altitude layer network cooperation method based on low-altitude airspace operation risk according to claim 1, characterized in that, Multiple optimized network parameter thresholds are obtained through optimization processing for UAV cooperative control, including: Continue to randomly set multiple network parameter thresholds for UAV flight at multiple altitude levels, and perform iterative optimization. After convergence, obtain multiple network parameter thresholds corresponding to the maximum cooperative fitness, and use them as multiple optimized network parameter thresholds for UAV cooperative control.

6. A multi-altitude layer network collaborative system for unmanned aerial vehicles (UAVs) based on low-altitude airspace operational risks, characterized in that: The system is used to implement the UAV multi-altitude layer network cooperative method based on low-altitude airspace operation risks as described in any one of claims 1-5, the system comprising: The low-altitude image layering module is used to acquire images within the low-altitude airspace, divide them into multiple altitude layers, identify the density of flying objects, and obtain the density of flying objects at multiple altitude layers. The signal strength impact analysis module is used to acquire multiple signal strengths for UAV control at multiple altitude levels, and to perform impact analysis and adjustment on multiple signal strengths based on the density of flying objects at multiple altitude levels to obtain multiple impact signal strengths; The network parameter evaluation module is used to randomly set multiple network parameter thresholds for UAV flight at multiple altitude levels, and calculate multiple safety factors and multiple control factors, including: Multiple network parameter thresholds are randomly set for UAV flight at multiple altitude levels, including a signal-to-noise ratio threshold. Obtain the average network parameter threshold set over a past period; Calculate the ratio of multiple network parameter thresholds to the average network parameter threshold to obtain multiple security factors; Calculate the ratio of the average network parameter threshold and multiple network parameter thresholds to obtain multiple manipulation coefficients; The collaborative optimization decision-making module is used to configure multiple safety weights and control weights based on the density of flying objects at multiple altitude levels and the strength of multiple influencing signals. It combines multiple safety coefficients and multiple control coefficients to calculate the collaborative fitness, and optimizes these to obtain multiple optimized network parameter thresholds for collaborative UAV control. This includes: Calculate the ratio of multiple object densities to a preset object density, and use these ratios as multiple density coefficients; Calculate the ratio of the preset signal strength to multiple influencing signal strengths, and use these ratios as multiple signal coefficients; Based on multiple density coefficients and multiple signal coefficients, multiple security weights and multiple control weights are calculated and assigned. The cooperative fitness degree is obtained by weighting multiple safety coefficients and multiple control coefficients according to the multiple safety weights and multiple control weights.

Images