Unmanned aerial vehicle cluster night operation safety monitoring method based on infrared thermal imaging

Abstract

This invention provides a method for safety monitoring of nighttime operations by drone swarms based on infrared thermal imaging, relating to the field of drone control technology. The method includes the following steps: collecting standard preset characteristic data and real-time flight characteristic data of the drone swarm during nighttime operations; preprocessing the collected real-time flight characteristic data to calculate the operational deviation coefficient; collecting environmental parameters of the flight scenario; establishing a multi-factor weighted formula to obtain environmental interference values; obtaining a continuous operational safety index; setting operational safety indicators; and determining the comprehensive risk result. This invention achieves quantitative early warning and graded decision-making for risks in swarm operations under complex dynamic environments. It overcomes the limitations of traditional monitoring methods that mainly rely on threshold alarms or single-parameter judgments, enabling more accurate identification of potential hazards under low-visibility nighttime conditions and supporting intelligent response strategies such as adaptive deceleration and emergency return-to-home, significantly improving the safety and task adaptability of drone swarm nighttime operations.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) control technology, specifically to a method for safety monitoring of UAV swarm operations at night based on infrared thermal imaging. Background Technology

[0002] In nighttime drone swarm operations, due to low visibility and complex, variable environments, relying solely on single-dimensional monitoring is insufficient to comprehensively assess operational risks. Therefore, it is crucial to consider both changes in the drone's own condition and environmental interference factors. On the one hand, deviations in the drone's own condition directly reflect a decline in flight control accuracy and mission execution capability; on the other hand, external environmental factors may exacerbate flight instability and the risk of sensor failure. These two factors are coupled and jointly affect operational safety. Only by incorporating both types of factors into safety monitoring methods can potential risks be more accurately identified, operational feasibility assessed, and the safe and efficient operation of drone swarms in complex nighttime environments ensured.

[0003] In the prior art, CN109270949A discloses a UAV flight control system, which includes: a processor module, a formation cluster remote control flight module, a formation cluster control module, a cruise control module, a return-to-home control module, a safety monitoring module, a ground control module, a smart charging module, and a memory. The return-to-home control module and the safety monitoring module are both connected to a communication antenna. The data transmission module and the data communication module are connected to the ground control module. The processor module is connected to both the ultrasonic ranging module and the image acquisition module. This solution obtains the current position and flight attitude, determines the flight area, and controls the UAV flight according to the flight control strategy, improving the efficiency and safety of formation flight, monitoring UAV malfunctions, and ensuring that faulty UAVs return to home in a timely manner.

[0004] However, the aforementioned existing technologies have shortcomings such as a single monitoring dimension and a lack of joint quantitative analysis of the environment and flight status. They often rely solely on a single flight parameter threshold for alarms, making it difficult to comprehensively assess the coupled impact of multiple environmental factors such as wind speed, fog, and ground reflection on the nighttime operations of the cluster. At the same time, their risk judgment logic often adopts a simple "intervention upon exceeding the standard" mode, lacking a graded early warning and adaptive adjustment mechanism. This makes it difficult to accurately identify potential risks in complex dynamic environments and to achieve intelligent graded decision-making. Consequently, false alarms or missed alarms are prone to occur in low visibility or high interference scenarios at night, resulting in relatively insufficient overall safety and mission adaptability.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a safety monitoring method for nighttime operations of drone swarms based on infrared thermal imaging, thereby addressing the problems mentioned in the background section. This invention achieves quantitative early warning and tiered decision-making for risks in swarm operations under complex dynamic environments. It overcomes the limitations of traditional monitoring methods that primarily rely on threshold alarms or single-parameter judgments, enabling more accurate identification of potential hazards under low-visibility nighttime conditions. Furthermore, it supports intelligent response strategies such as adaptive speed reduction and emergency return-to-home, significantly improving the safety and mission adaptability of drone swarm nighttime operations.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A method for safety monitoring of nighttime drone swarm operations based on infrared thermal imaging includes the following steps:

[0009] S1: Collect standard preset characteristic data of the UAV swarm during nighttime operations. The standard preset characteristic data includes: preset flight heading angle, preset waypoint coordinate sequence, and preset average swarm formation spacing under the UAV swarm design state; at the same time, collect real-time flight characteristic data of the UAVs. The real-time flight characteristic data includes: real-time flight heading angle, real-time waypoint coordinate sequence, and real-time average swarm formation spacing.

[0010] S2: Preprocess the collected real-time flight feature data. The preprocessing includes timestamp alignment, outlier filtering, and coordinate system normalization. Calculate the operation deviation coefficient based on the standard preset feature data and the real-time flight feature data to quantify the difference between the real-time flight state and the preset standard state.

[0011] S3: Collect flight scene environmental parameters, including ambient wind speed, obstacle density index, surface material reflectivity, and haze optical thickness.

[0012] S4: Establish a multi-factor weighting formula, normalize the flight scenario environmental parameters collected in S3 and input them into the multi-factor weighting formula, and obtain the environmental interference value after comprehensive calculation;

[0013] S5: Based on the operation deviation coefficient obtained from S2, the environmental disturbance value is coupled and processed to obtain the continuous operation safety index. The operation safety index is set. By comparing the continuous operation safety index with the operation safety index, the comprehensive risk result is determined, and a decision plan on whether to continue night operation is output.

[0014] Furthermore, in S1, the preset flight heading angle is pre-set based on the mission planning and is used to characterize the ideal flight direction under the design state. The real-time flight heading angle is the flight direction angle actually monitored during the operation and is used to reflect the current flight state.

[0015] The preset waypoint coordinate sequence is predefined based on the mission design and includes the starting point, turning point and ending point. The real-time waypoint coordinate sequence is dynamically collected through flight monitoring and records the actual trajectory.

[0016] The preset average spacing of the cluster formation is the ideal value under the design state, while the real-time average spacing of the cluster formation is the average distance measured during operation, reflecting the actual compactness of the formation.

[0017] Furthermore, during the preprocessing of outlier filtering in S2, if the value of any single data point in the collected real-time flight feature data... A value is considered an outlier if it meets the following conditions:

[0018]

[0019] in:

[0020] x represents the value of a single data point, referring to a measurement value in the real-time flight characteristic data;

[0021] μ is the average value of the parameter dataset in which the data point is located, and the parameter dataset includes the real-time flight heading angle dataset, the real-time waypoint coordinate sequence dataset, and the real-time cluster formation average spacing dataset.

[0022] k is a constant threshold coefficient used to define the boundary range of outliers;

[0023] σ is the sample standard deviation of the parameter dataset in which the data point is located, used to quantify the dispersion of the data;

[0024] And the formula for obtaining the sample mean μ is: Where n is the total number of data points, This represents the value of a single data point with index i.

[0025] Furthermore, the work deviation coefficient is calculated using the following formula:

[0026]

[0027] in:

[0028] DC is the work deviation coefficient;

[0029] To preset the flight heading angle, For real-time flight heading angle;

[0030] This is a preset sequence of waypoint coordinates; This is a real-time waypoint coordinate sequence;

[0031] The preset average spacing between cluster formations; This represents the average spacing between real-time cluster formations.

[0032] absolute deviation Quantify the degree of course deviation. Waypoint sequence deviation is defined as the average distance between corresponding points in the waypoint coordinate sequence; absolute deviation. Quantify changes in formation compactness.

[0033] Furthermore, the waypoint sequence deviation The calculation formula is:

[0034]

[0035] in:

[0036] m is the total number of waypoint sequences;

[0037] The x-axis value at the t-th coordinate point of the preset waypoint coordinate sequence; . represents the x-axis value at the t-th coordinate point in the real-time waypoint coordinate sequence;

[0038] The y-axis value at the t-th coordinate point of the preset waypoint coordinate sequence; The y-axis value at the t-th coordinate point in the real-time waypoint coordinate sequence;

[0039] via waypoint sequence deviation Quantify the path accuracy deviation during drone operations to identify the risk of flight path errors.

[0040] Furthermore, in step S3, the collected flight scenario environmental parameters are verified and cleaned. Outliers are filtered out using logical rules, and missing values ​​are imputed using a sliding window mean. The timestamps of each parameter are unified to the UAV cluster's operational clock reference. Spatial coordinates are mapped using a geographic information system to ensure that parameters such as wind speed and obstacle density match the UAV's real-time location. Finally, unit normalization is performed, and the original values ​​are linearly scaled based on a preset threshold.

[0041] The normalized ambient wind speed is obtained by dividing the ambient wind speed by the maximum operable wind speed threshold.

[0042] The obstacle density index is converted to a [0,1] index based on the region grid ratio to obtain a normalized obstacle density index;

[0043] Surface reflectance is converted to decimal form to obtain normalized surface material reflectance;

[0044] Normalized haze optical thickness was obtained by segmenting haze optical thickness according to visibility level.

[0045] Furthermore, the multi-factor weighting formula in S4 is as follows:

[0046]

[0047] in:

[0048] EI represents the environmental interference value;

[0049] WS is the normalized ambient wind speed, used to quantify the impact of wind speed on the stability of the drone;

[0050] OD stands for Normalized Obstacle Density Index, used to quantify the interference of obstacles on the flight path;

[0051] SR is the normalized surface reflectance, used to quantify the effect of surface reflection on UAV optical sensing.

[0052] HT is the normalized haze optical thickness, used to quantify the interference of haze on visibility and sensor accuracy;

[0053] These are the weighting coefficients for ambient wind speed, obstacle density index, surface material reflectivity, and haze optical thickness, respectively, and the weighting coefficients satisfy the following: , .

[0054] Furthermore, this also includes the weighting coefficients. The selection process is as follows: Based on historical operation data and expert experience, the impact of environmental parameters on the safety of drone night operations is prioritized, and the interference contribution of each parameter is quantified through risk event regression analysis; the initial weights are adjusted based on the characteristics of the current operation scenario, and the relative importance between parameters is calculated using the analytic hierarchy process to generate normalized weight values; the effectiveness of the weights is verified through real-time monitoring data, and periodic iterative optimization is performed to ensure that the weight allocation conforms to the dynamic environmental risk characteristics.

[0055] Furthermore, the formula upon which the coupling process is based is: SI stands for Continuous Operation Safety Index. It is a natural constant used to prevent the denominator from being zero.

[0056] Furthermore, the OSI (On-Site Safety Index) is set as the operational safety indicator, and the continuous operational safety index is compared with the operational safety indicator:

[0057] When SI≥ The risk level is determined to be low, and the operational status is close to the preset ideal conditions. Control the drone to continue the operation.

[0058] when <SI< At this point, there is an acceptable deviation or interference, and monitoring needs to be strengthened. The drone speed should be reduced by 20% to 50%, and the monitoring frequency should be increased.

[0059] When SI≤ If a serious deviation occurs or an extreme environment is encountered, the drone should be controlled to terminate its operation, return to base urgently, and the drone swarm should be controlled to autonomously avoid obstacles.

[0060] Compared with the prior art, the beneficial effects of the present invention are:

[0061] This invention comprehensively assesses risks by combining flight status and environmental conditions, and is specifically optimized for complex scenarios such as nighttime operations. It quantifies the deviation of the drone swarm's flight status from preset standards using an operational deviation coefficient. It also incorporates environmental interference values ​​to comprehensively consider the impact of multiple external factors such as wind speed, obstacles, surface reflectivity, and fog on nighttime operations. Finally, a continuous operational safety index is generated through a coupled model, enabling quantitative early warning and tiered decision-making for swarm operations in complex dynamic environments. This invention overcomes the limitations of traditional monitoring methods that rely primarily on threshold alarms or single-parameter judgments, enabling more accurate identification of potential hazards under low-visibility nighttime conditions. It also supports intelligent response strategies such as adaptive deceleration and emergency return-to-home, significantly improving the safety and mission adaptability of drone swarm operations at night. Attached Figure Description

[0062] Figure 1 This is a flowchart illustrating the safety monitoring method for nighttime operations of unmanned aerial vehicle (UAV) swarms based on infrared thermal imaging, as described in this invention. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0064] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0065] Example:

[0066] Please see Figure 1 The present invention provides the following technical solutions:

[0067] A method for safety monitoring of nighttime drone swarm operations based on infrared thermal imaging includes the following steps:

[0068] S1: Collect standard preset characteristic data of the UAV swarm during nighttime operations. The standard preset characteristic data includes: preset flight heading angle, preset waypoint coordinate sequence, and preset average swarm formation spacing under the UAV swarm design state; at the same time, collect real-time flight characteristic data of the UAVs. The real-time flight characteristic data includes: real-time flight heading angle, real-time waypoint coordinate sequence, and real-time average swarm formation spacing.

[0069] The preset flight heading angle is an ideal flight direction pre-set during the mission planning phase, taking into account factors such as mission objectives, geographical environment, airspace restrictions, and weather conditions. It provides a standard directional reference for the UAV throughout the nighttime operation and is an important reference for subsequent judgment of whether the heading has deviated. The real-time flight heading angle, on the other hand, is the current actual flight direction monitored and calculated in real time by sensors such as the inertial navigation system and magnetic compass on the UAV during the operation. It can accurately reflect the instantaneous heading changes of the UAV after being affected by external factors such as wind shear and turbulence.

[0070] The preset waypoint coordinate sequence is a set of coordinate points pre-planned on a geographic information system according to mission design requirements. These points typically include the start point, key turning points, and final destination. Each coordinate point corresponds to a spatial location that the UAV should traverse, forming a complete theoretical flight path to ensure that the UAV can efficiently and safely execute the mission along the predetermined route. The real-time waypoint coordinate sequence, on the other hand, is the actual coordinate points traversed during flight, dynamically collected and recorded by the UAV's onboard GPS or other positioning equipment. It reflects the actual flight trajectory of the UAV in complex environments and can visually display the differences between the real-time and preset paths.

[0071] The preset average spacing of the swarm formation is an ideal spacing value set during the mission planning phase based on the number of UAVs, the capacity of the operating airspace, and the requirements for cooperative collision avoidance. This value aims to ensure that the swarm can maintain efficient cooperative operation capabilities while having sufficient safety redundancy during nighttime flight. The real-time average spacing of the swarm formation is the average distance between each UAV measured and calculated in real time during actual operation through inter-UAV communication and positioning systems. This value can dynamically reflect the actual compactness of the formation under the influence of airflow disturbances, navigation errors, or external interference.

[0072] S2: Preprocess the collected real-time flight feature data. This preprocessing process first uses timestamp alignment to unify flight data from different sensors or data sources to the same time reference, ensuring consistency in the time dimension for subsequent comparisons and analyses.

[0073] Secondly, outlier filtering is performed, using statistical methods to identify and remove abnormal data points caused by factors such as sensor malfunctions and signal interference, thereby ensuring data quality. Finally, coordinate system unification is completed, converting waypoint coordinates from different reference frames to the same spatial coordinate system to eliminate calculation errors caused by coordinate differences. Based on preprocessed standard preset feature data and real-time flight feature data, an operational deviation coefficient is further calculated. This coefficient accurately quantifies the difference between the real-time flight state and the preset standard state by comprehensively comparing the deviation of key parameters such as heading angle, waypoint coordinates, and formation spacing, providing a reliable basis for subsequent safety assessments.

[0074] During the preprocessing for outlier filtering, any real-time flight feature data collected is considered an outlier if it meets the following conditions:

[0075]

[0076] in:

[0077] x represents the value of a single data point, referring to a measurement in the real-time flight characteristic data, and its absolute deviation from the mean μ. The fundamental basis for anomaly detection is: the greater the deviation, the higher the degree to which the data point deviates from the overall central tendency, and the greater the likelihood that it will be classified as an outlier. It is positively correlated with the probability of anomalies;

[0078] μ is the average value of the parameter dataset where the data point is located. The parameter dataset includes the real-time flight heading angle dataset, the real-time waypoint coordinate sequence dataset, and the real-time cluster formation average spacing dataset. When the data distribution is stable, μ can be used as the central reference of normal values. The deviation of x from μ is the key indicator for judging anomalies.

[0079] k is a constant threshold coefficient used to define the boundary range of outliers. k is directly proportional to the threshold value: the larger the k value, the more outliers are allowed. The wider the scope, the stricter the anomaly detection.

[0080] σ represents the sample standard deviation of the parameter dataset to which the data point is located. It quantifies the dispersion of the data. σ, together with the constant k, constitutes the threshold boundary for judging anomalies. A larger σ indicates higher data dispersion, at which point a higher allowable deviation is considered. The upper limit also increases accordingly, meaning that a larger absolute deviation is required for a data point to be considered an anomaly;

[0081] And the formula for obtaining the sample mean μ is: Where n is the total number of data points, This represents the value of a single data point with index i.

[0082] The work deviation coefficient is calculated using the following formula:

[0083]

[0084] in:

[0085] DC is the operational deviation coefficient. The larger the value, the greater the deviation and the higher the risk.

[0086] The preset flight heading angle represents the ideal flight direction under the design state of the drone swarm. The real-time flight heading angle represents the actual flight direction monitored during the operation. The heading angle deviation directly reflects whether the UAV is flying in the predetermined direction. The larger the value, the more severely the UAV is affected by factors such as wind shear, navigation error or control interference, resulting in the deviation of the actual flight path from the expected direction.

[0087] This is a preset sequence of waypoint coordinates; The deviation is a real-time waypoint coordinate sequence and is positively correlated with DC: the greater the average distance, the more serious the path deviation, the more likely the UAV will cross obstacles or enter dangerous airspace, and the operational deviation coefficient will rise accordingly.

[0088] The preset average spacing between cluster formations; The deviation value is the average spacing of the real-time cluster formation. It is positively correlated with DC: the larger the deviation, the more the formation structure deviates from the design state, the more unstable the relative positional relationship between the UAVs, which in turn affects the safety and coordination of the overall formation and increases the operational deviation coefficient.

[0089] absolute deviation Quantify the degree of course deviation. Waypoint sequence deviation is defined as the average distance between corresponding points in the waypoint coordinate sequence; absolute deviation. Quantify changes in formation compactness.

[0090] waypoint sequence deviation The calculation formula is:

[0091]

[0092] in:

[0093] m is the total number of waypoint sequences, which determines the number of samples involved in the calculation. With a fixed sequence length, the contribution of each waypoint is averaged out. However, m itself does not directly drive the change in deviation, but rather serves as a normalization factor to ensure that the deviation value reflects the overall average degree of deviation.

[0094] The x-axis value at the t-th coordinate point of the preset waypoint coordinate sequence; The x-axis value of the real-time waypoint coordinate sequence at the t-th coordinate point; The y-axis value at the t-th coordinate point of the preset waypoint coordinate sequence; The y-axis value at the t-th coordinate point in the real-time waypoint coordinate sequence;

[0095] The deviation of each coordinate component is positively correlated with the final average distance deviation. That is, the increase in positional deviation in any direction will directly increase the path accuracy deviation value, thus more significantly reflecting the degree of spatial deviation between the actual trajectory of the UAV and the preset path.

[0096] via waypoint sequence deviation Quantify the path accuracy deviation during drone operations to identify the risk of flight path errors.

[0097] S3: Collect flight scene environmental parameters, including ambient wind speed, obstacle density index, surface material reflectivity, and haze optical thickness.

[0098] The collected flight scenario environmental parameters undergo data verification and cleaning. First, the raw data is checked for reasonableness using preset logical rules to remove obvious erroneous values ​​caused by momentary sensor failures or communication anomalies, ensuring that the data used in subsequent analysis has basic reliability. Then, a sliding window mean interpolation method is used to fill in missing values. This method calculates local means based on valid data from nearby time points, maintaining the continuity of environmental parameter changes while avoiding calculation interruptions due to data gaps.

[0099] After data cleaning, the timestamps of all environmental parameters were uniformly calibrated to the drone swarm's operational clock reference, ensuring strict synchronization of all parameters in the time dimension. Next, a geographic information system (GIS) was used to map the environmental parameters to spatial coordinates, enabling parameters such as wind speed and obstacle density to accurately match the drone's real-time location, achieving spatial correlation between environmental data and flight status. Finally, unit normalization was performed, linearly scaling the original values ​​according to preset thresholds for each environmental parameter, converting parameters of different dimensions into dimensionless relative values, laying the foundation for subsequent comprehensive calculation of environmental interference values.

[0100] Normalize the ambient wind speed: Divide the real-time collected ambient wind speed value by the preset maximum operable wind speed threshold to obtain a normalized ambient wind speed between 0 and 1. This value is used to measure the relative magnitude of the current wind speed with respect to the upper limit of safe operation of the drone, thereby intuitively reflecting the potential impact of wind speed on flight stability.

[0101] Normalize the obstacle density index: Based on the grid division of the area where the UAV is currently located, calculate the area ratio of obstacles in each grid and convert it into a normalized obstacle density index with a value between 0 and 1. This index can accurately characterize the density of obstacles in the area and their risk of interference to the flight path.

[0102] Normalize the reflectivity of the surface material: Convert the original surface reflectivity data into decimal form and standardize its value range to between 0 and 1. In this way, the impact of different surface types on the detection performance of UAV optical sensors can be compared on the same scale.

[0103] The optical thickness of haze is normalized: Based on the real-time monitored optical thickness values ​​of haze, and referring to the visibility level classification standard, it is segmented and mapped to a normalized haze optical thickness between 0 and 1. This value is used to quantify the degree of interference of haze on visibility and sensor accuracy. After the above normalization process, all environmental parameters are transformed into dimensionless relative indicators.

[0104] S4: Establish a multi-factor weighting formula, normalize the flight scenario environmental parameters collected in S3 and input them into the multi-factor weighting formula, and obtain the environmental interference value after comprehensive calculation;

[0105] The multi-factor weighting formula is:

[0106]

[0107] in:

[0108] EI represents the environmental interference value;

[0109] WS is the normalized ambient wind speed, converted into a relative value within the range of [0,1], used to quantify the degree of interference of the current wind speed on the stability of the UAV. The higher the wind speed, the more significant the impact of turbulence and crosswinds on the UAV during flight, which increases the difficulty of attitude control and energy consumption, thereby increasing the risk of instability or deviation from the flight path;

[0110] Obstacle density index (OD) is used to quantify the interference of obstacles on the flight path. The higher the obstacle density, the more potential collision points the UAV needs to avoid during flight, the heavier the path planning complexity and the burden on the obstacle avoidance system, and the significantly higher the probability of scraping or collision. Therefore, OD is positively correlated with obstacle density index (EI).

[0111] SR is the normalized reflectivity of the ground surface, used to quantify the impact of ground reflection on the optical sensing of drones. The higher the reflectivity, the stronger the reflection of infrared radiation by the ground surface, which may lead to glare, false hotspots, or decreased contrast in imaging, affecting the drone's ability to identify targets and obstacles. Therefore, SR is positively correlated with EI.

[0112] HT represents the normalized optical thickness of haze, used to quantify the interference of haze on visibility and sensor accuracy. The greater the haze optical thickness, the lower the atmospheric transparency, resulting in decreased detection range and image clarity for infrared thermal imaging. It can also cause ranging errors and target recognition delays, directly threatening flight safety. Therefore, HT is positively correlated with EI: the thicker the haze, the higher the normalized value, and the greater the environmental interference.

[0113] These are the weighting coefficients for ambient wind speed, obstacle density index, surface material reflectivity, and haze optical thickness, respectively.

[0114] Ambient wind speed Considered the most basic interference factor, the drone's own flight control system has a certain degree of wind resistance, therefore its weight is relatively low; the reflectivity of the ground surface material. The main factor affecting the detection accuracy of infrared thermal imaging is its interference level, which is higher than that of wind speed when relying on optical sensors at night. However, it is still secondary compared to factors that directly threaten flight path safety; Obstacle density index Directly related to the risk of drones colliding with their surroundings, the threat to safety is significantly amplified at night when visibility is limited; therefore, it has a higher weight than reflectivity. Haze optical thickness... The highest weight is assigned because smog not only severely weakens the detection range and image quality of infrared thermal imaging, but also directly affects the drone's ability to identify obstacles and terrain. The combined risk of reduced visibility is most critical during nighttime operations, hence the highest weight coefficient. Therefore, the weight coefficient satisfies: , .

[0115] For weighting coefficients The selection process involves: collecting environmental parameter data and corresponding safety incident cases recorded during historical nighttime operations; organizing an evaluation team composed of UAV flight experts, meteorologists, and environmental engineers; and combining expert experience with data analysis to initially determine the priority of the impact of environmental parameters such as wind speed, obstacle density, surface reflectivity, and haze optical thickness on the safety of UAV nighttime operations. Based on this, a risk event regression analysis model is used to correlate various safety events with real-time environmental parameters, quantifying the interference contribution of each parameter in the risk occurrence process, thus providing an objective basis for initial weight setting.

[0116] Based on the specific characteristics of the current operational scenario, this embodiment includes the geographical environment, meteorological conditions, and flight mission requirements of the operational area. The initial weights are adaptively adjusted to better reflect the risk distribution in the actual environment. Using the analytic hierarchy process (AHP), experts compare the relative importance of each parameter pairwise to construct a judgment matrix and calculate the weight coefficients for each parameter. After consistency verification, normalized weight values ​​are generated to ensure a scientific and reasonable weight allocation.

[0117] During the operation, real-time monitoring data is used to verify the effectiveness of the weight allocation, observe the degree of consistency between the safety assessment results and the actual flight status, and regularly iterate and optimize the weights based on newly accumulated data, so that the weight system can dynamically follow the changes in environmental risk characteristics and always ensure the accurate quantification of safety interference to night operations.

[0118] S5: Based on the operation deviation coefficient obtained from S2, the environmental interference value is coupled and processed. The deviation coefficient, which reflects the degree of deviation of the UAV's own flight state, is combined with the environmental interference value, which characterizes the degree of influence of the external environment on flight safety. By comprehensively evaluating the combined effect of internal and external factors on operation safety, a continuous operation safety index is calculated. This index is used to quantify the comprehensive safety level of the UAV continuing to perform its mission under the current complex conditions.

[0119] Based on task requirements and safety standards, a work safety indicator is set as a benchmark threshold. A multi-level comparison is then performed between the continuously calculated work safety index and this indicator. The overall risk result is determined based on the different ranges the index falls into, including low risk, medium risk, and high risk levels. Finally, a clear decision-making plan is output based on the risk assessment results.

[0120] The formula upon which the coupling process is based is: SI stands for Continuous Operation Safety Index. It is a natural constant used to prevent the denominator from being zero.

[0121] Set the operational safety index as OSI, and compare the continuous operational safety index with the operational safety index:

[0122] When SI≥ The risk level is determined to be low, and the operational status is close to the preset ideal conditions. Control the drone to continue the operation.

[0123] when <SI< At this point, there is an acceptable deviation or interference, and monitoring needs to be strengthened. The drone speed should be reduced by 20% to 50%, and the monitoring frequency should be increased.

[0124] When SI≤ If a serious deviation occurs or an extreme environment is encountered, the drone should be controlled to terminate its operation, return to base urgently, and the drone swarm should be controlled to autonomously avoid obstacles.

[0125] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0126] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0127] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0128] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for safety monitoring of nighttime operations of unmanned aerial vehicle (UAV) swarms based on infrared thermal imaging, characterized in that, Includes the following steps: S1: Collect standard preset characteristic data of the UAV swarm during nighttime operations. The standard preset characteristic data includes: preset flight heading angle, preset waypoint coordinate sequence, and preset average swarm formation spacing under the UAV swarm design state; at the same time, collect real-time flight characteristic data of the UAVs. The real-time flight characteristic data includes: real-time flight heading angle, real-time waypoint coordinate sequence, and real-time average swarm formation spacing. S2: Preprocess the collected real-time flight feature data. The preprocessing includes timestamp alignment, outlier filtering, and coordinate system normalization. Calculate the operation deviation coefficient based on the standard preset feature data and the real-time flight feature data to quantify the difference between the real-time flight state and the preset standard state. S3: Collect flight scene environmental parameters, including ambient wind speed, obstacle density index, surface material reflectivity, and haze optical thickness. S4: Establish a multi-factor weighting formula, normalize the flight scenario environmental parameters collected in S3 and input them into the multi-factor weighting formula, and obtain the environmental interference value after comprehensive calculation; S5: Based on the operation deviation coefficient obtained from S2, the environmental disturbance value is coupled and processed to obtain the continuous operation safety index. The operation safety index is set. By comparing the continuous operation safety index with the operation safety index, the comprehensive risk result is determined, and a decision plan on whether to continue night operation is output.

2. The method for safety monitoring of nighttime operations of unmanned aerial vehicle (UAV) swarms based on infrared thermal imaging according to claim 1, characterized in that: In S1, the preset flight heading angle is pre-set based on the mission planning and is used to characterize the ideal flight direction under the design state. The real-time flight heading angle is the flight direction angle actually monitored during the operation and is used to reflect the current flight state. The preset waypoint coordinate sequence is predefined based on the mission design and includes the starting point, turning point and ending point. The real-time waypoint coordinate sequence is dynamically collected through flight monitoring and records the actual trajectory. The preset average spacing of the cluster formation is the ideal value under the design state, while the real-time average spacing of the cluster formation is the average distance measured during operation, reflecting the actual compactness of the formation.

3. The method for safety monitoring of nighttime operations of UAV swarms based on infrared thermal imaging according to claim 1, characterized in that: During the preprocessing of outlier filtering in S2, if the value of any single data point in the collected real-time flight feature data... A value is considered an outlier if it meets the following conditions: ； in: The value of a single data point refers to a measurement in the real-time flight characteristic data; It is the average value of the parameter dataset where the data point is located, and the parameter dataset includes the real-time flight heading angle dataset, the real-time waypoint coordinate sequence dataset, and the real-time cluster formation average spacing dataset. This is a constant threshold coefficient used to define the boundary range of outliers; The standard deviation of the parameter dataset to which the data point is located is used to quantify the dispersion of the data. And the sample mean The formula for obtaining it is: Where n is the total number of data points, This represents the value of a single data point with index i.

4. The method for safety monitoring of nighttime operations of UAV swarms based on infrared thermal imaging according to claim 3, characterized in that: The work deviation coefficient is calculated using the following formula: ； in: This is the work deviation coefficient; To preset the flight heading angle, For real-time flight heading angle; This is a preset sequence of waypoint coordinates; This is a real-time waypoint coordinate sequence; The preset average spacing between cluster formations; This represents the average spacing between real-time cluster formations. absolute deviation Quantify the degree of course deviation. Waypoint sequence deviation is defined as the average distance between corresponding points in the waypoint coordinate sequence; absolute deviation. Quantify changes in formation compactness.

5. The method for safety monitoring of nighttime operations of unmanned aerial vehicle (UAV) swarms based on infrared thermal imaging according to claim 4, characterized in that: waypoint sequence deviation The calculation formula is: ； in: m is the total number of waypoint sequences; The x-axis value at the t-th coordinate point of the preset waypoint coordinate sequence; The x-axis value of the real-time waypoint coordinate sequence at the t-th coordinate point; The y-axis value at the t-th coordinate point of the preset waypoint coordinate sequence; The y-axis value at the t-th coordinate point in the real-time waypoint coordinate sequence; via waypoint sequence deviation Quantify the path accuracy deviation during drone operations to identify the risk of flight path errors.

6. The method for safety monitoring of nighttime operations of unmanned aerial vehicle (UAV) swarms based on infrared thermal imaging according to claim 1, characterized in that: In step S3, the collected flight scenario environmental parameters are verified and cleaned. Outliers are filtered out using logical rules, and missing values ​​are imputed using a sliding window mean. The timestamps of each parameter are unified to the UAV cluster's operational clock reference. Spatial coordinates are mapped using a geographic information system to ensure that parameters such as wind speed and obstacle density match the real-time location of the UAV. Finally, unit normalization is performed, and the original values ​​are linearly scaled based on a preset threshold. The normalized ambient wind speed is obtained by dividing the ambient wind speed by the maximum operable wind speed threshold. The obstacle density index is converted to a [0,1] index based on the region grid ratio to obtain a normalized obstacle density index; Surface reflectance is converted to decimal form to obtain normalized surface material reflectance; Normalized haze optical thickness was obtained by segmenting haze optical thickness according to visibility level.

7. The method for safety monitoring of nighttime operations of unmanned aerial vehicle (UAV) swarms based on infrared thermal imaging according to claim 6, characterized in that: The multi-factor weighting formula in S4 is as follows: ； in: EI represents the environmental interference value; WS is the normalized ambient wind speed, used to quantify the impact of wind speed on the stability of the drone; OD stands for Normalized Obstacle Density Index, used to quantify the interference of obstacles on the flight path; SR is the normalized surface reflectance, used to quantify the effect of surface reflection on UAV optical sensing. HT is the normalized haze optical thickness, used to quantify the interference of haze on visibility and sensor accuracy; Do not assign weights to ambient wind speed, obstacle density index, surface material reflectivity, and haze optical thickness, and these weights must satisfy the following conditions: , .

8. The method for safety monitoring of nighttime operations of UAV swarms based on infrared thermal imaging according to claim 7, characterized in that: It also includes the weighting coefficients The selection process is as follows: Based on historical operation data and expert experience, the impact of environmental parameters on the safety of UAV nighttime operations is prioritized, and the interference contribution of each parameter is quantified through risk event regression analysis. The initial weights are adjusted based on the characteristics of the current work scenario, and the relative importance between parameters is calculated using the analytic hierarchy process to generate normalized weight values. The effectiveness of the weights is verified by real-time monitoring data, and periodic iterative optimization is performed to ensure that the weight allocation conforms to the risk characteristics of the dynamic environment.

9. The method for safety monitoring of nighttime operations of unmanned aerial vehicle (UAV) swarms based on infrared thermal imaging according to claim 7, characterized in that: The formula upon which the coupling process is based is: SI stands for Continuous Operation Safety Index. It is a natural constant used to prevent the denominator from being zero.

10. The method for safety monitoring of nighttime operations of unmanned aerial vehicle (UAV) swarms based on infrared thermal imaging according to claim 9, characterized in that: Set the operational safety index as OSI, and compare the continuous operational safety index with the operational safety index: When SI≥ The risk level is determined to be low, and the operational status is close to the preset ideal conditions. Control the drone to continue the operation. when <SI< At this point, there is an acceptable deviation or interference, and monitoring needs to be strengthened. The drone speed should be reduced by 20% to 50%, and the monitoring frequency should be increased. When SI≤ If a serious deviation occurs or an extreme environment is encountered, the drone should be controlled to terminate its operation, return to base urgently, and the drone swarm should be controlled to autonomously avoid obstacles.

Images