Intelligent pod control method and system for unmanned aerial vehicle

Abstract

The present application provides a method and system for controlling an intelligent pod of a UAV. The method obtains multiple frames of thermal radiation images through the infrared imaging module of the UAV pod, generates a dynamic temperature field map through heat source decoupling, and uses a graph neural network to analyze the spatial correlation between heat sources. Based on the temperature gradient vector, a damping torque matching the wind disturbance is generated, the filtering parameters of the magnetorheological pan-tilt platform are optimized, and the interference intensity between energy transfer paths is calibrated. The threat propagation path is generated in combination with the thermal diffusion characteristics, and the pod deflection angle is dynamically adjusted to synchronize the center of the optical field of view with the threat center of mass, thereby achieving precise tracking and anti-interference control. Used for intelligent pod control of UAVs. The present application realizes precise and stable tracking of threat targets by the UAV pod through intelligent heat source analysis and adaptive adjustment.

Description

Technical Field

[0001] The present application relates to the field of intelligent pod control technology, and in particular to an intelligent pod control method and system for unmanned aerial vehicles. Background Art

[0002] In nighttime security scenarios, detecting illegal drone swarm intrusions presents complex challenges: rapidly identifying multiple heat source targets in low-light environments, overcoming environmental interference such as wind disturbance, and tracking dynamic threats. Traditional optoelectronic methods struggle to effectively distinguish dense heat sources from background noise, requiring intelligent pods to possess high-precision thermal radiation analysis capabilities, adaptive anti-interference control, and swarm target trajectory prediction.

[0003] The current mainstream solution uses "infrared imaging + deep learning" technology, extracting thermal image features through convolutional neural networks, combining it with Kalman filtering to predict target motion trajectories, and suppressing gimbal vibrations using fixed-threshold filtering. For example, a patent proposes using the YOLOv5 model to detect heat source targets and using PID control to adjust the pod's field of view tracking.

[0004] This solution inadequately addresses the energy transfer interference between heat sources and the dynamic wind disturbance coupling effects: first, CNN has difficulty modeling the complex spatial correlations between heat sources, resulting in a high false detection rate for dense targets; second, fixed parameter filtering cannot adapt to changes in the wind disturbance spectrum, resulting in a delayed gimbal response; third, the impact of heat diffusion paths on threat propagation is not considered, resulting in significant trajectory prediction deviations. Summary of the Invention

[0005] The present application provides a method and system for controlling an intelligent pod for a UAV, so as to solve the problem of poor control effect of the intelligent pod in the prior art.

[0006] In a first aspect, the present application provides a method for controlling an intelligent pod of a drone, comprising:

[0007] The infrared imaging module in the intelligent pod acquires multiple frames of thermal radiation images of the drone, performs heat source decoupling processing on the multiple frames of thermal radiation images, and generates a dynamic temperature field map of each heat source;

[0008] Based on the temperature gradient vectors of the dynamic temperature field maps of all the heat sources, a graph neural network is used to analyze the spatial correlation between the heat sources;

[0009] Based on the spatial correlation between heat sources, a damping torque matching the wind disturbance spectrum is generated, and the filtering parameters of the magnetic fluid damping pan / tilt platform in the intelligent pod are modified according to the frequency response characteristics of the damping torque;

[0010] Mapping the temperature gradient vector to a corresponding energy transfer path, and using the filtering parameter to calibrate the interference intensity between two adjacent energy transfer paths;

[0011] extracting a heat diffusion feature from the temperature gradient vector, and generating a threat propagation path based on the calibrated interference intensity and the heat diffusion feature;

[0012] According to the threat propagation path, the deflection angle of the smart pod is dynamically planned so that the center of the optical field of view is spatially synchronized with the geometric center of mass of the threat propagation path.

[0013] Optionally, extracting a heat diffusion feature from the temperature gradient vector and generating a threat propagation path according to the calibrated interference intensity and the heat diffusion feature includes:

[0014] The temperature gradient vectors in the dynamic temperature field map of each heat source are decomposed into time series, and multiple features of the temperature gradient direction at continuous time points, such as the change trajectory, fluctuation pattern, and cumulative change amount, are extracted to form a direction change sequence, an intensity fluctuation sequence, and a heat diffusion energy value. Specifically, the direction change sequence is formed based on the change trajectory of the temperature gradient direction at continuous time points; the intensity fluctuation sequence is formed based on the fluctuation pattern of the temperature gradient magnitude at continuous time points; and the heat diffusion energy value is formed based on the cumulative change amount of the temperature gradient vector within a preset time window.

[0015] combining the direction change sequence, the intensity fluctuation sequence and the heat diffusion energy value into a heat diffusion characteristic vector for each heat source;

[0016] Based on the calibrated interference intensity, an interference intensity table between heat sources is constructed, where each table entry stores the interference intensity value between a pair of heat sources, and the characteristic similarity of the heat diffusion feature vectors of any two heat sources is calculated. The similarity is determined by fusing the trajectory repeatability and fluctuation consistency of the two heat sources.

[0017] Multiplying the feature similarity with the interference intensity value of the corresponding heat source pair in the interference intensity table to obtain the propagation correlation weight between the heat sources;

[0018] Taking the heat source with the largest heat diffusion energy value as the starting node, select the adjacent heat source with the highest propagation association weight with the current heat source from the heat sources that have not been traversed. Add the selected heat source to the path sequence and update it as the current heat source. Repeat the above steps until the propagation association weight is lower than the preset threshold or there are no optional adjacent heat sources. Output the path sequence as the threat propagation path.

[0019] Optionally, dynamically planning a deflection angle of the smart pod according to the threat propagation path so that the center of the optical field of view is spatially synchronized with the geometric center of mass of the threat propagation path includes:

[0020] Obtaining the spatial coordinates of all heat sources in the threat propagation path to form a coordinate set, and calculating the geometric centroid of the coordinate set, wherein the geometric centroid includes the centroid coordinates of multiple components;

[0021] Get the current pointing angle of the smart pod, where the current pointing angle includes the horizontal yaw angle and the vertical pitch angle;

[0022] The center of mass coordinates are converted to the pod coordinate system through coordinate transformation. The horizontal yaw angle correction and vertical pitch angle correction are calculated based on the converted center of mass coordinates and the spatial relative position of the smart pod, and used as the target pointing angle;

[0023] The difference between the current pointing angle and the target pointing angle is decomposed into angular velocity components. Kinematic constraints are applied to the angular velocity components based on the inertial parameters of the magnetofluid damping gimbal. A continuous angle change sequence that satisfies the constraints is output, and the smart pod is controlled to deflect according to the continuous angle change sequence so that the center of the optical field of view is aligned with the geometric center of mass coordinates.

[0024] Optionally, a multi-frame thermal radiation image of the UAV is acquired by an infrared imaging module in the intelligent pod, and heat source decoupling processing is performed on the multi-frame thermal radiation image to generate a dynamic temperature field map of each heat source, including:

[0025] The infrared imaging module of the intelligent pod captures multiple frames of thermal radiation images of the drone at consecutive time points. Each frame of thermal radiation image contains the temperature values ​​of multiple pixel points.

[0026] Tracking the heat source profile of each independent heat source in consecutive frames based on the radiation intensity distribution and spatial position stability of the multiple frames of thermal radiation images;

[0027] Determining the radiation contribution of each independent heat source in the multiple frames of thermal radiation images using the tracked heat source profile;

[0028] According to the radiation contribution, a unique identifier is assigned to each heat source, and its temperature distribution data in each frame of the image is extracted; the temperature distribution data of each heat source is integrated in time series to generate a dynamic temperature field map of the heat source.

[0029] Optionally, based on the temperature gradient vectors of the dynamic temperature field maps of all the heat sources, a graph neural network is used to analyze the spatial correlation between the heat sources, including:

[0030] Extract the temperature gradient vector from the dynamic temperature field map of each heat source;

[0031] Inputting the temperature gradient vector as a node feature into a graph neural network, wherein the graph neural network learns the dependency relationship between the node features;

[0032] Through iterative calculation of the graph neural network, the association weights between nodes are output, and the association weights represent the degree of mutual influence of temperature changes between heat sources;

[0033] A spatial correlation description between heat sources is generated according to the correlation weights.

[0034] Optionally, generating a damping torque that matches the wind disturbance spectrum based on the spatial correlation between heat sources and correcting filtering parameters of a magnetic fluid damping pan / tilt platform in the intelligent pod based on the frequency response characteristics of the damping torque includes:

[0035] Based on the spatial correlation between the heat sources, a spatial relationship model is constructed, wherein the spatial relationship model represents the position of each heat source and the degree of mutual influence as a connection weight;

[0036] Simulating the dynamic response of the wind disturbance spectrum at different frequencies through the spatial relationship model to convert the connection weights into moment generation factors, and calculating the moment amplitude and phase at multiple frequency points based on the moment generation factors;

[0037] synthesizing a damping torque sequence according to the torque amplitude and phase, wherein a frequency spectrum characteristic of the damping torque sequence is consistent with a characteristic of the wind disturbance frequency spectrum;

[0038] Comparing the amplitude response and phase response of the extracted frequency response characteristics of the damping torque sequence with current filtering parameters of the magnetic fluid damping pan / tilt platform in the intelligent pod to generate a parameter deviation;

[0039] According to the parameter deviation, the cutoff frequency and the gain coefficient in the current filtering parameters are adjusted to generate modified filtering parameters.

[0040] Optionally, mapping the temperature gradient vector to a corresponding energy transfer path, and using the filtering parameter to calibrate the interference intensity between two adjacent energy transfer paths includes:

[0041] Taking each vector element in the temperature gradient vector as a starting point, performing path tracing along the vector direction to generate multiple energy transfer paths;

[0042] calculating an overlapping area between two adjacent energy transfer paths, and quantifying the overlapping area as an initial interference intensity;

[0043] The initial interference intensity is scaled and adjusted using the cutoff frequency and gain coefficient in the filtering parameters to determine the calibrated interference intensity, wherein the scaling and adjustment process includes: using the cutoff frequency to filter out high-frequency interference components, and using the gain coefficient to amplify or attenuate the amplitude of the initial interference intensity.

[0044] In a second aspect, the present application provides an intelligent pod control system for a drone, comprising:

[0045] The first generation module is used to obtain multiple frames of thermal radiation images of the drone through the infrared imaging module in the intelligent pod, perform heat source decoupling processing on the multiple frames of thermal radiation images, and generate a dynamic temperature field map of each heat source;

[0046] An analysis module for analyzing the spatial correlation between heat sources using a graph neural network based on the temperature gradient vectors of the dynamic temperature field maps of all the heat sources;

[0047] A correction module is used to generate a damping torque that matches the wind disturbance spectrum based on the spatial correlation between heat sources, and to correct the filtering parameters of the magnetic fluid damping pan / tilt platform in the intelligent pod based on the frequency response characteristics of the damping torque;

[0048] a calibration module, configured to map the temperature gradient vector to a corresponding energy transfer path, and to calibrate the interference intensity between two adjacent energy transfer paths using the filtering parameters;

[0049] a second generating module, configured to extract a heat diffusion feature from the temperature gradient vector and generate a threat propagation path according to the calibrated interference intensity and the heat diffusion feature;

[0050] A planning module is used to dynamically plan the deflection angle of the smart pod according to the threat propagation path so that the center of the optical field of view is spatially synchronized with the geometric center of mass of the threat propagation path.

[0051] In an embodiment of the present application, multi-frame thermal radiation images of the drone are acquired through an infrared imaging module in an intelligent pod, and heat source decoupling processing is performed on the multi-frame thermal radiation images to generate a dynamic temperature field map of each heat source; based on the temperature gradient vectors of the dynamic temperature field maps of all the heat sources, a graph neural network is used to analyze the spatial correlation between the heat sources; based on the spatial correlation between the heat sources, a damping torque matching the wind disturbance spectrum is generated, and the filtering parameters of the magnetorheological damping gimbal in the intelligent pod are corrected according to the frequency response characteristics of the damping torque; the temperature gradient vector is mapped to a corresponding energy transfer path, and the interference intensity between two adjacent energy transfer paths is calibrated using the filtering parameters; the heat diffusion feature is extracted from the temperature gradient vector, and a threat propagation path is generated based on the calibrated interference intensity and the heat diffusion feature; based on the threat propagation path, the deflection angle of the intelligent pod is dynamically planned to keep the center of the optical field of view spatially synchronized with the geometric center of mass of the threat propagation path.

[0052] The technical solution of this application has the following beneficial effects:

[0053] The system uses multi-frame thermal radiation images to separate independent heat sources, eliminating background noise interference and providing high-precision thermodynamic data for subsequent analysis. It quantifies the interaction between temperature gradient vectors between heat sources, resolving the problem of false detection of densely packed targets and improving the accuracy of target cluster recognition. It adaptively generates a frequency-matched damping torque to suppress gimbal vibration caused by wind disturbances and enhance dynamic stability. It optimizes the calculation of inter-path interference intensity through filtering parameters to reduce trajectory prediction errors caused by cross-effects of thermal radiation. It integrates thermal diffusion characteristics with calibrated interference intensity to predict threat diffusion trends and enable active trajectory tracking. It synchronizes the optical field of view with the threat center of mass to ensure continuous target lock and improve tracking response speed and accuracy.

[0054] Furthermore, this application extracts the directional changes, fluctuation patterns, and energy accumulation characteristics of the temperature gradient vector through time series decomposition to construct a heat diffusion feature vector. The application then calculates the propagation correlation weight by combining the interference intensity table with feature similarity, generating a threat propagation path originating from the maximum energy heat source. The geometric center of mass is then calculated based on the coordinates of the heat source along the path. Coordinate transformation and kinematic constraints are then used to generate a pod deflection sequence, achieving spatial synchronization between the optical field of view and the center of mass.

[0055] This application accurately predicts the evolution trend of threat paths through the fusion of thermal diffusion spatiotemporal features and dynamic calculation of propagation weights; combined with center of mass tracking and kinematic constraint control, the pod can quickly and stably align with the threat core, significantly improving the tracking robustness and anti-interference capability of cluster targets.

[0056] These and other aspects of the present application will become more readily apparent from the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0058] Figure 1 A flow chart of an intelligent pod control method for a drone provided by the present application is shown;

[0059] Figure 2 The figure shows a schematic structural diagram of an intelligent pod control system for a drone provided by the present application;

[0060] Figure 3 A schematic structural diagram of a computing device provided by the present application is shown. DETAILED DESCRIPTION

[0061] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0062] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this document or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any order of execution. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to being different types.

[0063] In nighttime drone swarm monitoring scenarios, existing approaches based on "infrared imaging + deep learning" suffer from significant limitations: First, traditional convolutional neural networks struggle to effectively model the dynamic energy transfer relationships between heat sources, resulting in a high false detection rate for densely packed heat source targets. Second, fixed-parameter filtering algorithms are unable to adapt to the time-varying spectral characteristics of wind disturbances, causing response lag and tracking jitter in the magnetohydrodynamic damping gimbal. Third, existing methods lack quantitative analysis of heat diffusion paths, leading to systematic biases in threat propagation trajectory predictions, severely impacting the effectiveness of coordinated tracking of swarm targets. These shortcomings stem from the existing technology's fragmented approach to thermodynamic correlations, environmental disturbance coupling, and energy propagation mechanisms.

[0064] To address these issues, the present invention proposes an intelligent pod control method based on dynamic heat source coupling analysis. This method uses a graph neural network to construct a spatial correlation model of temperature gradient vectors and incorporates an adaptive damping adjustment mechanism based on wind disturbance spectrum matching to achieve high-precision decoupling and stable tracking of heat source targets. Specifically, heat source decoupling is performed on multiple frames of thermal radiation images to generate a dynamic temperature field map. The spatiotemporal characteristics of the temperature gradient vectors are then used to construct energy transfer paths and calibrate interference intensity. Finally, a threat propagation path is generated and the pod's deflection angle is dynamically planned. This method innovatively integrates thermodynamic property analysis, environmental disturbance suppression, and target motion prediction. The graph neural network addresses the problem of false detection of dense heat sources, while the frequency response of magnetic fluid damping is used to correct gimbal instability caused by wind disturbances. Accurate reconstruction of threat propagation paths is achieved based on heat diffusion characteristics and interference intensity calibration. Compared to existing technologies, this solution improves the tracking accuracy of clustered targets in complex nighttime environments while reducing the amplitude of field-of-view jitter caused by wind disturbances, significantly enhancing the reliability and adaptability of drone pods in security monitoring scenarios.

[0065] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0066] Figure 1 The present invention provides a flow chart of an intelligent pod control method for a drone, as shown in FIG. Figure 1 As shown, the method includes:

[0067] 101. Acquire multiple frames of thermal radiation images of the drone through the infrared imaging module in the intelligent pod, perform heat source decoupling processing on the multiple frames of thermal radiation images, and generate a dynamic temperature field map of each heat source.

[0068] Optionally, step 101 may specifically include:

[0069] 1011. Using the infrared imaging module of the intelligent pod, multiple frames of thermal radiation images of the drone are captured at consecutive time points, each frame of the thermal radiation image containing temperature values ​​of multiple pixel points;

[0070] 1012. Tracking the heat source profile of each independent heat source in consecutive frames based on the radiation intensity distribution and spatial position stability of the multiple frames of thermal radiation images;

[0071] 1013. Determine the radiation contribution of each independent heat source in the multiple frames of thermal radiation images using the tracked heat source profile;

[0072] In the above scheme, the thermal radiation image refers to the two-dimensional temperature distribution data captured by the infrared imaging module, where each pixel value corresponds to the radiation intensity at a certain point in the scene. Heat source decoupling refers to the process of separating the radiation data of each independent heat source from overlapping thermal radiation images containing multiple heat sources. The heat source contour refers to the boundary of the continuous area in the thermal radiation image where the temperature is significantly higher than the background, and is a closed polygon formed by adjacent high-temperature pixels. The radiation contribution refers to the temperature influence weight of a single heat source in a mixed pixel. The true temperature value of each heat source is separated through decoupling calculation. The dynamic temperature field map contains the distribution of radiation gradient vectors caused by engine power fluctuations on the target surface.

[0073] In the embodiment of the present application, first, step 1011 is used to continuously capture the thermal radiation energy of the drone surface using the infrared imaging module in the smart pod at a fixed time interval, such as 0.1 seconds / frame; the detector is composed of tens of thousands of independent pixels, such as a 384×288 array to form physical pixel points, each pixel receives infrared radiation photons in the target area and converts them into electrical signal intensity values ​​through photoelectric conversion; the signal is quantized by the analog-to-digital converter, combined with the preset material emissivity, such as = such as motor aluminum alloy ε = 0.88 and the ambient temperature compensation parameter, according to the formula , calculate the absolute temperature value of each pixel, where S is the electrical signal strength, k is the detector calibration coefficient, and σ is the Stefan constant; finally, the system maps the pixel spatial position into a two-dimensional temperature matrix, directly generating a multi-frame thermal radiation image containing 110,592, or 384×288, pixel temperature values, and repeating this process at consecutive time points to obtain a thermal radiation image containing temperature values ​​of multiple pixels.

[0074] Secondly, step 1012 is used to calculate the gradient amplitude of a single-frame thermal image based on the radiation intensity distribution of multiple-frame thermal radiation images, and extract the temperature mutation area as a candidate heat source; then, a heat source centroid motion model is established based on the spatial position stability of the frame thermal radiation image. According to this model, when the displacement threshold is less than the maximum allowable displacement threshold and the variance of the centroid movement direction is less than 15° within three consecutive frames, it is judged to be a stable heat source; the heat source contour that meets the stability condition is optimally matched through the algorithm to achieve trajectory locking and tracking of the same heat source between multiple frames.

[0075] Finally, in step 1013, based on the traced heat source contours, the scan line algorithm is used to fill the heat source polygon boundary for each heat source to generate a binary mask matrix; then the radiation value of the heat source is directly extracted in the non-overlapping area. When multiple heat source contours overlap, the radiation contribution is allocated according to the physical characteristics.

[0076] For example, in a nighttime urban security scenario, the pod monitors an invading drone cluster in area A.

[0077] Specifically, the infrared module captured five consecutive thermal image frames with a 0.1-second interval between frames. Three moving heat sources, designated as heat sources 1-3, and one static background heat source, designated as heat source 4, were detected. Optical flow analysis identified heat source 1 as shifting by only 2 pixels between consecutive frames, meeting the stability threshold, and its outline consisting of 50 boundary pixels. NMF decomposition revealed that the temperature of a mixed pixel, 82°C at coordinates [120, 300], was contributed by 65% ​​(weight 0.65) to heat source 1 and 35% (weight 0.35) to heat source 2. The final output was a dynamic map of heat source 1: the temperature increased from 78°C to 85°C in 2 seconds, with a spatial gradient of 1.2°C per pixel.

[0078] This step achieves precise separation and dynamic modeling of heat sources through joint spatiotemporal analysis. Displacement stability is used to eliminate background interference, and matrix decomposition is combined to quantify the radiative contribution of each heat source within the mixed pixel, generating a high-fidelity spatiotemporal evolution map of the temperature field. This approach eliminates the effects of overlapping dense heat sources and provides reliable independent heat source dynamic data for subsequent correlation analysis.

[0079] 102. Analyze the spatial correlation between the heat sources using a graph neural network based on the temperature gradient vectors of the dynamic temperature field maps of all the heat sources;

[0080] Optionally, step 102 may specifically include:

[0081] 1021. Extracting the temperature gradient vector from the dynamic temperature field map of each heat source;

[0082] 1022. Input the temperature gradient vector as a node feature into a graph neural network, where the graph neural network learns the dependency relationship between the node features.

[0083] 1023. Outputting association weights between nodes through iterative calculations of the graph neural network, wherein the association weights represent the degree of mutual influence of temperature changes between heat sources;

[0084] 1024. Generate a spatial correlation description between heat sources based on the correlation weights.

[0085] In the above solution, the dynamic temperature field map refers to the time-series temperature distribution data generated for each heat source, including spatial location and temperature value matrices. The temperature gradient vector is a physical quantity that describes the direction and intensity of temperature changes in the temperature field, consisting of a directional component (angle) and a magnitude component (rate of change). A graph neural network is a deep learning model used to process graph-structured data and can learn relationships between nodes. Node features are attribute information vectors for each node (i.e., heat source) in the graph. Association weights refer to the degree of mutual influence of temperature changes between heat sources. Spatial association descriptions refer to the results of the interaction analysis, presented in the form of a heat source relationship diagram and a text report.

[0086] In the embodiment of the present application, firstly, the dynamic temperature field map of each heat source is spatially differentiated in step 1021, that is, the horizontal gradient is T ( x +1, y ) T ( x 1, y ), the vertical gradient is T ( x , y +1) T (x , y 1); where T(x,y) is the temperature value at position (x,y), the horizontal gradient is the temperature change rate in the x direction, and the vertical gradient is the temperature change rate in the y direction. Then calculate the gradient amplitude and angle of each heat source, that is, , ,in is the temperature change intensity, is the direction of temperature change, is the vertical temperature gradient component, is the horizontal temperature gradient component. The above calculation results are output as the temperature gradient vector of each heat source.

[0087] Secondly, through step 1022, the temperature gradient vector of each heat source, that is, the average gradient amplitude, the standard deviation of the gradient change, the main gradient direction angle and the spatial position coordinates are first encoded together as a node feature vector as the input layer of the neural network; then, the graph neural network uses a multi-layer message passing mechanism to enable each node to collect the feature information of its neighboring nodes, and perform nonlinear transformation and feature fusion operations in the hidden layer. This process allows the network to automatically capture the heat conduction pattern between heat sources. For example, when the main gradient direction of a heat source continues to point to another heat source, the network will strengthen the connection weight between the two nodes; finally, the node state is iteratively updated through multi-layer stacked graph convolution layers, so that the network gradually learns the dependency relationship of the temperature change transmission law characteristics between heat sources, such as the radiation effect of high-temperature heat sources on adjacent low-temperature heat sources, the synchronous temperature change trend between heat sources of the same material, and other deep dependencies.

[0088] Next, in step 1023, in the initial iteration layer of the graph neural network, each heat source node transmits its temperature gradient characteristics, including change intensity and direction information, to adjacent nodes through fully connected edges. The neural network analyzes the similarity of temperature change patterns through a feature comparison algorithm. For example, when the gradient directions of the main motor and battery nodes continuously maintain a difference of less than 15°, a preliminary correlation signal is generated. Then, an attention screening mechanism is introduced in the intermediate iteration layer to focus on analyzing physical connections, such as heat-conducting structures or spatial proximity, such as pairs of heat source nodes with a distance of less than 20 cm. Effective connections are strengthened through temporal correlation analysis of temperature changes. If the temperature rise and fall changes of two heat sources show a correlation coefficient of more than 0.8 in the time series, their association weight is significantly increased. Finally, in the output layer, the accumulated association evidence is processed through a nonlinear activation function, and a continuous weight value in the interval [0, 1] is assigned to each pair of heat source nodes, such as 0.86 to indicate strong mutual influence. This weight accurately quantifies the transmission efficiency of the temperature rise of heat source A leading to the temperature change of heat source B, completing the mathematical characterization of the degree of mutual temperature influence.

[0089] Finally, through step 1024, the numerical association weights are first mapped into a visual thermal matrix diagram, the matrix coordinate axes are sorted according to the physical positions of the heat sources, and the mutual influence relationship between the heat sources is intuitively displayed with light and dark colors, for example, dark red represents strong influence and light blue represents weak influence; at the same time, a dynamic relationship network diagram is generated, in which the node size is proportional to the sum of the weight output of the heat source, the edge width is proportional to the association weight value, and the core heat conduction path is marked; finally, a multi-dimensional spatial correlation description is automatically generated based on the weight threshold, and three types of key correlations are identified, namely, the dominant heat source with the highest output weight, the strongly coupled heat source pair with a two-way weight greater than 0.8, and the isolated heat source with a total correlation degree less than 0.3.

[0090] For example, a drone contains three heat sources: the main motor (node ​​A), the battery (node ​​B), and the controller (node ​​C).

[0091] Specifically, temperature gradient extraction revealed that the dominant gradient of A was directed toward B (angle 120°), while the gradient of C was uniformly distributed (σ = 0.15). An initial fully connected graph was then constructed using a graph neural network. After training, the output correlation weights were 0.82 (strong correlation), 0.31 (weak correlation), and 0.19 (very weak correlation). The resulting spatial correlation description indicated that the main motor had a significant thermal impact on the battery, recommending the addition of a thermal barrier between the two. The controller was relatively independent and required no adjustment.

[0092] This method automatically discovers the implicit correlation between heat sources in complex systems, reveals the temperature propagation path, and helps optimize thermal management design; the generated correlation description intuitively displays the key influencing sources and affected targets, providing a decision-making basis for preventing overheating failures; the adaptive learning ability of graph neural networks effectively improves the accuracy of thermal correlation analysis.

[0093] 103. Generate a damping torque that matches the wind disturbance spectrum based on the spatial correlation between the heat sources, and modify the filtering parameters of the magnetic fluid damping pan / tilt platform in the intelligent pod based on the frequency response characteristics of the damping torque;

[0094] Optionally, step 103 may specifically include:

[0095] 1031. Based on the spatial correlation between the heat sources, construct a spatial relationship model, wherein the spatial relationship model represents the position of each heat source and the degree of mutual influence as a connection weight;

[0096] 1032. Simulate the dynamic response of the wind disturbance spectrum at different frequencies using the spatial relationship model to convert the connection weights into torque generation factors, and calculate the torque amplitudes and phases at multiple frequency points based on the torque generation factors;

[0097] 1033. Synthesize a damping torque sequence according to the torque amplitude and phase, wherein the frequency spectrum characteristics of the damping torque sequence are consistent with the characteristics of the wind disturbance spectrum;

[0098] 1034. Compare the amplitude response and phase response of the extracted frequency response characteristics of the damping torque sequence with current filtering parameters of the magnetic fluid damping pan / tilt platform in the intelligent pod to generate a parameter deviation;

[0099] 1035. Adjust the cutoff frequency and gain coefficient in the current filtering parameters according to the parameter deviation to generate corrected filtering parameters.

[0100] In the above scheme, the spatial relationship model refers to the position of each heat source and the degree of mutual influence represented as connection weights. The torque generation factor refers to the conversion coefficient that converts the heat source correlation strength into a physical and mechanical response. It maps the dimensionless correlation weight into a dimensional force that can act on the gimbal. The damping torque refers to the direction deviation of the optical axis of the smart pod under a 12m / s side wind disturbance is less than the preset threshold. The damping torque sequence refers to a set of torque values ​​that changes with time. Its fluctuation characteristics are completely matched with the external wind disturbance and are used to actively offset the influence of wind vibration. The frequency response characteristics refer to the response ability of the magnetofluid damping gimbal to vibrations of different frequencies, including response speed and suppression effect. The filtering parameter refers to the core tuning parameter in the magnetofluid damping control system, which determines which frequencies of vibration interference the system filters.

[0101] In this embodiment of the present application, step 1031 first obtains the precise position coordinates (x, y, z) of each heat source in three-dimensional space based on the smart pod coordinate system, and maps these positions to nodes in the spatial network. Next, the mutual influence value calculated in step 102 is directly converted into a connection weight value between the nodes. A dynamic connection edge generation mechanism is then established. When the distance between two heat source nodes is less than a preset physical threshold, such as 10 cm, and the mutual influence degree is greater than 0.4, a connection edge with mechanical properties is generated between them. The line width and color depth of the edge are proportional to the connection weight, forming an intuitive spatial relationship model. For example, a weight of 0.8 generates an 8-pixel wide red solid line, and a weight of 0.3 generates a 3-pixel wide blue dashed line.

[0102] Next, in step 1032, the spatial model is simulated for wind disturbance spectrum response. The 0-100 Hz wind spectrum is divided into 100 frequency points. The model vibration response is analyzed at the 50 Hz frequency point. Based on the weight values ​​of the connecting edges and the relationship between the heat source positions, the sensitivity distribution of different frequency bands is determined to convert the associated weights into torque generation factors. For example, a weight of 0.7-1.0 corresponds to a high-frequency factor of 1.0, 0.4-0.7 corresponds to a mid-frequency factor of 0.6, and less than 0.4 corresponds to a low-frequency factor of 0.3. The specific torque amplitude is then calculated by multiplying the generation factor by a reference coefficient, where the reference coefficient is obtained from the wind tunnel test data table. The phase angle is calculated based on the node spacing, which adds 0.3° phase lag per centimeter. For example, with a node spacing of 10 cm and a weight of 0.85, the output amplitude at 50 Hz is 0.25 Nm and the phase is -30°.

[0103] Next, the frequency band is finely complemented according to the energy distribution characteristics of the torque amplitude, phase and wind disturbance spectrum through step 1033, and every 1Hz interpolation point is added in the 30-50Hz core area where the wind energy is concentrated. The amplitude and phase of the newly added point are calculated by the linear interpolation algorithm. For example, the 33Hz amplitude is the average of 32Hz and 34Hz, and the phase is the nearest neighbor value. Then, a full-band torque vector table is constructed to ensure that each frequency point has accurately corresponding amplitude and phase data. Then, time domain waveform reconstruction is performed, and the sinusoidal torque components of each frequency point are superimposed and calculated according to the time axis, that is, 1ms is used. The time step is t = 0ms, and the sum of the initial values ​​of all sine waves is calculated, for example, -0.2Nm. At t = 1ms, the superposition of all sine waves after 1ms evolution is calculated, for example, +0.05Nm. The 1000ms damping torque sequence is generated by continuous iteration. Finally, spectrum calibration is performed, and the synthesized sequence is analyzed by fast Fourier transform and compared with the original wind spectrum curve. The frequency bands with deviations of >3% are fine-tuned in amplitude and corrected in phase offset until the main peak frequency deviation is <0.5Hz and the harmonic distortion rate is <1%. The damping torque sequence consistent with the spectral characteristics of the wind disturbance is output.

[0104] Then, in step 1034, the key characteristic points of the frequency response characteristic curve of the synthesized damping torque sequence are recorded, such as the amplitude value of the main frequency peak 35dB@35Hz, the phase lag angle 155° and the full-band response trend; at the same time, the filter parameter set currently in effect for the magnetic fluid damping pan / tilt is read, such as the center frequency 40Hz, bandwidth ±5Hz, gain 1.2, and phase compensation 15°; then, a multi-dimensional comparison is performed in the three-coordinate analyzer, wherein when comparing the amplitude response, the amplitude value 35dB of the damping sequence at 35Hz is detected to be consistent with the actual response of the pan / tilt filter at this frequency. 33dB, that is, the difference is +2dB, to generate the amplitude deviation matrix; when performing phase response comparison, the hysteresis difference between the damping sequence phase of 155° and the pan-tilt preset 150° is detected at the same frequency point of 35Hz to form a phase deviation spectrum; finally, the two sets of data are integrated through the intelligent calibration algorithm to extract core deviations, such as the main frequency amplitude difference of +2dB, the main frequency phase difference of +5°, and the bandwidth deviation of -3Hz, and comprehensively generate structural parameter deviations, such as the amplitude compensation coefficient of +0.1, the center frequency shift of +3Hz, the phase compensation supplement of +5°, and the bandwidth expansion coefficient of ×1.2.

[0105] Finally, in step 1035, a hierarchical correction protocol is initiated based on the parameter deviation for four core parameters: amplitude compensation coefficient, center frequency shift, phase compensation supplement, and bandwidth expansion coefficient. The amplitude compensation coefficient is input into the digital gain controller, triggering the gain coefficient to be increased in three steps from the current value of 1.0 to 1.15, thereby adjusting the gain coefficient. The center frequency offset is then repositioned by the numerically controlled oscillator, for example, by adding a 2Hz offset from the original 40Hz to lock it at 42Hz, thereby adjusting the cutoff frequency. Phase lag is then injected into the delay circuit for time compensation, and bandwidth expansion is achieved by adjusting the capacitor array of the bandpass filter. Hardware-in-the-loop verification is then performed, inputting the original wind disturbance spectrum onto a simulated wind vibration platform and monitoring the fluctuation range of the corrected gimbal attitude angle. If the residual vibration energy exceeds a threshold, a second-level fine-tuning is initiated. Finally, a timestamped encrypted control instruction set is generated and written to the FPGA controller of the magnetofluid damping gimbal via the dual CAN bus. The original parameters are simultaneously backed up to a secure storage area to generate the corrected filter parameters.

[0106] For example, in a night inspection mission of brand A drone at location B.

[0107] Specifically, the smart pod detects the spatial correlation weight of 0.88 between the main motor heat source C (coordinates (0.1, 0.2, 0.3)) and the infrared lens driver heat source D (coordinates (0.15, 0.25, 0.35)). First, an elastic model connecting the two heat sources is constructed, such as an 8.8px orange connecting line with a spacing of 18mm <20cm threshold. After thermal conduction verification, the driver temperature rises by 12°C within 3 seconds after the motor temperature rises by 20°C. When a sudden gust of 6 is encountered, the system converts the weight into a torque factor, i.e., 0.92, and calculates the torque amplitude at 28Hz. The damping torque sequence was synthesized. The frequency response characteristics of the sequence, 28 Hz @ 29.5 dB and a phase angle of 48°, were extracted and compared with the current parameters of the magnetohydrodynamic gimbal to generate parameter deviations of +1.3 dB amplitude deviation and -1.9 Hz center frequency deviation. Based on this, the gimbal filter parameters were modified, with the cutoff frequency shifted from 28-38 Hz to 26.1-36.1 Hz and the gain coefficient adjusted from 1.1 to 1.23. Ultimately, the gimbal jitter was reduced from ±1.2° to ±0.4°, ensuring the pod's high-definition imaging capabilities in strong winds.

[0108] This solution constructs a physical model by quantifying the spatial correlation of heat sources and converting the thermal coupling strength into anti-wind disturbance control parameters. Based on the dynamic response mechanism, a damping torque sequence with precise spectrum matching is generated to achieve mechanical equivalent reproduction of external wind disturbances. Through the closed-loop feedback system, the frequency response characteristics are intelligently compared to drive the magnetofluid damping gimbal to dynamically adjust the cutoff frequency and gain coefficient. Ultimately, an adaptive mapping of the spatial relationship of heat sources to vibration reduction control is achieved, significantly improving the pod's imaging stability and target tracking capabilities in complex wind disturbance environments, while optimizing the system's response timeliness and avoiding manual parameter adjustment errors.

[0109] 104. Map the temperature gradient vector to a corresponding energy transfer path, and use the filtering parameter to calibrate the interference intensity between two adjacent energy transfer paths;

[0110] Optionally, step 104 may specifically include:

[0111] 1041. Taking each vector element in the temperature gradient vector as a starting point, performing path tracing along the vector direction to generate multiple energy transfer paths;

[0112] 1042. Calculate the overlapping area between two adjacent energy transfer paths, and quantify the overlapping area as the initial interference intensity;

[0113] 1043. Use the cutoff frequency and gain coefficient in the filtering parameters to scale the initial interference intensity to determine the calibrated interference intensity, wherein the scaling adjustment process includes: using the cutoff frequency to filter out high-frequency interference components, and using the gain coefficient to amplify or attenuate the amplitude of the initial interference intensity.

[0114] In the above scheme, the temperature gradient vector refers to a physical quantity that describes the direction and magnitude of temperature changes on the surface of a heat source, and contains information in two dimensions: angle (radiation direction) and intensity (rate of change). The energy transfer path refers to a straight path extending from the center of the heat source along the direction of the temperature gradient vector, representing the main heat diffusion channel. The initial interference intensity is the quantified value of the superposition of thermal energy in the intersection area of ​​adjacent energy paths, reflecting the degree of energy crosstalk between the paths. The cutoff frequency is the highest frequency threshold (in Hz) allowed by the filtering system to pass, which is used to block high-frequency interference. The gain coefficient is a dimensionless signal amplitude control factor used to amplify or attenuate the interference intensity. The calibrated interference intensity is the quantified interference value after cutoff frequency filtering and gain coefficient adjustment, representing the actual effective thermal interference. The overlap area is determined based on the minimum distance between the spatial points and the angle between the path directions.

[0115] In an embodiment of the present application, first, a set of temperature gradient vectors of each heat source surface is obtained through step 1041, that is, the direction angle and intensity value, with the starting coordinate point of the vector as the path reference point; then, a straight line is traced along the vector direction, using a ray stepping algorithm starting from the reference point, with each advancement step equal to the characteristic size of the heat source material, and the temperature change rate is monitored during the advancement process. Tracking is stopped when the temperature gradient decays to 30% of the initial value; then, a sequence of path trajectory points is recorded in three-dimensional space coordinates, and paths are automatically branched for complex heat sources, such as splitting three secondary paths in the vector direction for an irregular heat sink heat source; finally, all paths are integrated to generate multiple energy transfer paths.

[0116] Secondly, according to the energy transfer path generated in step 1041, a spatial scanning algorithm is used to detect all adjacent path pairs whose spacing is less than the average size of the heat source through step 1042; then, the intersection coordinates are calculated for the path pairs that meet the conditions, that is, the intersection is determined by solving the analytical solution of the linear equations of the two paths, and the nearest projection point is calculated when there is no intersection; then, an overlapping area model with the intersection as the core of the ellipse is constructed, and the length of the major axis of the ellipse is set to 0.5 times the spacing between the two paths, the minor axis is 0.2 times the spacing, and the main axis direction is parallel to the bisector of the angle between the two paths; then, all temperature data points in the elliptical area are extracted from the dynamic temperature field map, and the Gaussian weighted integral algorithm is used to calculate the thermal field energy: the temperature value of each pixel point is multiplied by the coordinate weight coefficient, and the total value of the regional thermal energy is accumulated; finally, the thermal energy value is mapped to the dimensionless initial interference intensity through the thermodynamic-electrical conversion coefficient.

[0117] Finally, step 1043 obtains the initial interference intensity value calculated in step 1042, and simultaneously reads the cutoff frequency and gain coefficient of the currently effective filter parameters of the magnetic fluid damping pan / tilt platform. Next, high-frequency interference is identified based on the spatial angle characteristics of the two energy transfer paths. If the path angle is greater than 60 degrees, it is marked as a high-frequency interference component. For example, if path A is 45 degrees and path B is 120 degrees, the intersection angle is 75 degrees, which is greater than 60 degrees. Then, high-frequency filtering is performed based on the cutoff frequency. When the cutoff frequency is less than 30 Hz, 100% suppression is applied to the marked high-frequency interference component. When 30Hz≤cutoff frequency≤50Hz, linear proportional suppression is used. For example, 70% intensity is retained when the cutoff frequency = 40Hz. When the cutoff frequency is greater than 50Hz, the high-frequency interference component is completely retained. Then the gain coefficient is applied to adjust the amplitude. When the interference intensity is lower than the noise threshold, the 1.5th power of the gain coefficient is used for intensity compensation to suppress weak signal attenuation. When it is higher than the safety threshold, the 0.8th power of the gain coefficient is used for attenuation protection to avoid overload. Finally, the weight is re-adjusted according to the thermal conductivity of the path material to generate the calibrated interference intensity value.

[0118] For example, in the inspection operation of Model A drone, its intelligent pod detects the gradient vector of the motor heat source (45°, 15°C / cm) and the gradient vector of the controller heat source (120°, 8°C / cm).

[0119] Specifically, three motor paths (45° / 30° / 60° directions) and two controller paths (120° / 150° directions) are first generated. The intersection of the 45° and 120° paths is detected at the coordinates (0.22, 0.31), and the area of ​​the elliptical overlap region is calculated to be 0.8 cm², the average temperature is 72°C, and the initial interference intensity is 57.6°C·cm². The current filtering parameters of the gimbal are then read (cutoff frequency 35Hz, gain 1.25). Because the intersection angle is 75°>60°, 80% of the high-frequency components are retained. After gain adjustment, the output calibration intensity is 1.25×0.8×57.6×0.85≈49°C·cm², providing an accurate interference quantification basis for thermal management decisions.

[0120] This method achieves a deep integration of the physical properties of the temperature field and signal processing technology, converting the temperature gradient vector into an operational energy transfer path; accurately capturing path cross-thermal interference through geometric modeling; dynamically calibrating the interference intensity using gimbal filter parameters to effectively distinguish between real thermal coupling and high-frequency noise; and ultimately improving the pod system's ability to quantitatively analyze heat conduction interference, providing a physically consistent decision-making benchmark for thermal management strategy optimization, and significantly enhancing the pod's adaptability in complex thermal environments.

[0121] 105. Extracting a heat diffusion feature from the temperature gradient vector, and generating a threat propagation path according to the calibrated interference intensity and the heat diffusion feature;

[0122] Optionally, step 105 may specifically include:

[0123] 1051. Perform time series decomposition on the temperature gradient vectors in the dynamic temperature field map of each heat source, extract multiple features of the temperature gradient direction at consecutive time points, including the change trajectory, fluctuation pattern, and cumulative change amount, to form a direction change sequence, an intensity fluctuation sequence, and a heat diffusion energy value. Specifically, the direction change sequence is formed based on the change trajectory of the temperature gradient direction at consecutive time points; the intensity fluctuation sequence is formed based on the fluctuation pattern of the temperature gradient magnitude at consecutive time points; and the heat diffusion energy value is formed based on the cumulative change amount of the temperature gradient vector within a preset time window.

[0124] 1052. Combining the direction change sequence, the intensity fluctuation sequence, and the heat diffusion energy value into a heat diffusion characteristic vector for each heat source;

[0125] 1053. Construct an interference intensity table between heat sources based on the calibrated interference intensity, wherein each table entry stores the interference intensity value between a pair of heat sources, and calculate the feature similarity of the heat diffusion feature vectors of any two heat sources. The similarity is determined by fusing the trajectory repeatability and fluctuation consistency of the two heat sources.

[0126] 1054. Multiply the feature similarity by the interference intensity value of the corresponding heat source pair in the interference intensity table to obtain the propagation association weight between the heat sources;

[0127] 1055. Taking the heat source with the largest heat diffusion energy value as the starting node, select the adjacent heat source with the highest propagation association weight with the current heat source from the heat sources that have not been traversed, add the selected heat source to the path sequence, and update it as the current heat source. Repeat the above steps until the propagation association weight is lower than the preset threshold or there are no optional adjacent heat sources, and output the path sequence as the threat propagation path.

[0128] In the above scheme, the temperature gradient vector refers to a physical quantity that characterizes the direction and magnitude of the temperature change rate of the heat source surface, including angle (change direction) and intensity (change amount). The heat diffusion feature refers to the three key features extracted from the temperature gradient vector, namely the direction change sequence (trajectory time series change pattern), the intensity fluctuation sequence (gradient amplitude time series evolution), and the heat diffusion energy value (gradient change cumulative effect). The interference intensity table refers to a two-dimensional matrix that stores the interference intensity after calibration between heat source pairs, quantifying the cross-influence of heat sources. Feature similarity refers to the similarity index calculated by integrating trajectory overlap and fluctuation synchronization. The propagation correlation weight refers to the product of feature similarity and interference intensity, which determines the priority of thermal threat transmission. The threat propagation path refers to the sequence of conduction links generated along the highest correlation weight starting from the maximum heat source energy node. Trajectory repeatability and fluctuation consistency refer to the trajectory matching of the direction change sequences of two heat sources, the calculation of trajectory overlap, the phase alignment of the intensity fluctuation sequences of two heat sources, and the calculation of fluctuation consistency.

[0129] In the embodiment of the application, first, step 1051 performs time series decomposition on the temperature gradient vector in each frame of the atlas, that is, taking the vector data from the 1st second to the 10th second to form a vector sequence on the time axis, calculating the angular change value of the vector direction between two adjacent seconds, and continuously recording the change value of 9 time steps to form a direction change sequence, thereby extracting multiple features of the change trajectory, fluctuation pattern, and cumulative change amount of the temperature gradient direction at continuous time points; then, the temporal change of the gradient direction is analyzed, and the vector direction offset between adjacent frames is tracked, for example, 102° at the 1st second → 105° at the 2nd second → offset +3°, to generate a direction change sequence containing 9 time steps, and synchronously record the direction fluctuation entropy value, a high entropy value indicates directional disorder; at the same time, the gradient intensity evolution law is extracted, the standard deviation of the fluctuation amplitude reflected by the gradient of each frame, and the difference in the sudden change intensity valley captured by the peak of adjacent frames are calculated to form a two-parameter intensity fluctuation sequence; finally, within the 10-second time window, the thermal diffusion energy value is generated based on the linear accumulation of the product of the gradient intensity vector modulus and the temperature change rate, that is, the gradient intensity is integrated every second.

[0130] Next, step 1052 integrates the three independent features generated in step 1051: the direction change sequence, the intensity fluctuation sequence, and the heat diffusion energy value. These are then concatenated end-to-end in dimensional order, with the direction sequence placed in the first 9 positions of the vector, the intensity sequence in the 10th and 11th positions, and the energy value fixed at the 12th position, forming a 12-dimensional original feature vector. Principal component analysis (PCA) dimensionality reduction optimization is then performed to calculate the eigenvectors of the covariance matrix and retain the top five principal components with cumulative contributions exceeding 95%, such as PC1, the principal component of the direction time series pattern, and PC2, the principal component of the intensity fluctuation. This compresses the 12-dimensional vector into a 5-dimensional core feature. Finally, the heat source spatial location label is added to generate an 8-dimensional standardized feature vector. The entire process strictly adheres to the engineering path of "feature splicing → dimensionality compression → spatial anchoring" to ensure that the feature vector simultaneously contains the dynamic time series characteristics of heat diffusion, the energy intensity properties, and the eigenvectors of the physical location information.

[0131] Next, in step 1053, a heat source interference intensity matrix is ​​constructed based on the calibrated interference intensity values ​​output from step 1043. This matrix is ​​a lower triangular matrix with heat source IDs as row and column indices. Each entry stores the interference intensity value of a unique heat source pair, such as 0.78 for the main motor and infrared camera module, forming a relational network basis for thermal interference within the pod. Next, for all heat source pairs, feature similarity is calculated using a dual similarity fusion algorithm. The first step analyzes the trajectory repetition of the direction change sequence and uses dynamic time warping to compare the sequence morphology. For example, the waveform matching degree between the main motor sequence [+3°, +5°, -2°...] and the infrared camera sequence [+2°, +1°, +3°...] is 0.85. The second step evaluates the fluctuation consistency of the intensity fluctuation sequence and detects synchronous fluctuation patterns using the covariance index. For example, if both heat sources experience a sudden increase in intensity greater than 15% at 5 seconds, high covariance is indicated. Finally, a trajectory repetition weight of 0.6 and a fluctuation consistency weight of 0.4 are linearly fused to output a feature similarity in the range of 0.0-1.0.

[0132] Then, in step 1054, a two-factor coupling calculation is performed on the corresponding items of the feature similarity and interference intensity table according to the heat source calculated in step 1053, that is, a scalar multiplication of the feature similarity and the interference intensity value is performed, for example, 0.878 × 0.78 = 0.685, to obtain the propagation association weight;

[0133] Finally, through step 1055, the heat source with the largest heat diffusion energy value, such as the main motor with 185.3 energy units, is used as the path starting node and added to the threat path sequence. Then, all adjacent nodes with propagation association weights with the current heat source are retrieved from the untraversed heat source set, and the heat source with the highest weight is selected, such as the main motor → infrared camera with a weight of 0.806. Then, double verification is performed. If the weight is ≥ the preset threshold, the heat source is added to the path sequence and updated to the current node. With the updated node as the base point, its untraversed adjacent heat sources are repeatedly retrieved, and expansion is stopped when the weight is less than the threshold. Finally, the search is terminated when there are no reachable adjacent nodes or all weights are lower than the threshold, and an ordered threat propagation path sequence is output.

[0134] For example, in the smart pod system performing power inspection tasks.

[0135] Specifically, the time-series temperature gradient data of the three core heat sources (main motor module C, infrared camera module D, and image processor E) are first extracted to form a direction change sequence and intensity fluctuation pattern, and the thermal diffusion energy value of each heat source is calculated. Secondly, the feature similarity of module CD is constructed to 0.878, such as the comprehensive direction trajectory matching degree of 0.85 and the fluctuation consistency of 0.92, and the propagation weight of 0.685 is generated by combining the pre-stored interference intensity table. Then, starting from the maximum energy source module C, the propagation weights of adjacent heat sources are retrieved, and the module D with the highest weight is selected to join the path. Then, with module D as the current node, the weight of module E of 0.339 does not reach the threshold, and the path expansion is terminated. Finally, the threat propagation path [main motor C→infrared camera D] is output, which drives the pod cooling system to directional enhance the heat dissipation power of module D by 20%, while reducing the workload of module C by 15%, effectively blocking the transmission link of thermal threats within the system and ensuring the continuous and stable execution of inspection tasks.

[0136] This method achieves deep coupling of thermal physical properties and topological analysis, accurately captures the essence of heat conduction through multi-dimensional heat diffusion feature extraction; innovatively integrates interference intensity and feature similarity to generate propagation weights; reveals the thermal threat conduction path based on an energy-driven path search algorithm; and finally constructs a threat propagation model with strong interpretability and clear physical meaning, providing reliable decision-making support for UAV thermal safety management and significantly improving the system's forward-looking prediction capabilities for thermal failure risks.

[0137] 106. Based on the threat propagation path, dynamically plan the deflection angle of the smart pod so that the center of the optical field of view maintains spatial synchronization with the geometric center of mass of the threat propagation path.

[0138] Optionally, step 106 may specifically include:

[0139] 1061. Obtain the spatial coordinates of all heat sources in the threat propagation path to form a coordinate set, and calculate the geometric centroid of the coordinate set, where the geometric centroid includes the centroid coordinates of multiple components.

[0140] 1062. Obtain the current pointing angle of the intelligent pod, where the current pointing angle includes a horizontal yaw angle and a vertical pitch angle;

[0141] 1063. Convert the center of mass coordinates to the pod coordinate system through coordinate transformation, calculate the horizontal yaw angle correction and the vertical pitch angle correction based on the converted center of mass coordinates and the spatial relative position of the smart pod, and use them as the target pointing angle;

[0142] 1064. Decompose the difference between the current pointing angle and the target pointing angle into angular velocity components. Perform kinematic constraints on the angular velocity components based on the inertial parameters of the magnetofluid damping gimbal. Output a continuous angle change sequence that satisfies the constraints. Control the intelligent pod to deflect according to the continuous angle change sequence so that the center of the optical field of view is aligned with the geometric center of mass coordinates.

[0143] In the above scheme, the threat propagation path refers to the directional conduction channel of heat energy between components within the pod. This path consists of a sequence of heat source nodes (e.g., [motor → lens driver → controller]). Each node contains precise spatial coordinates. The edges connecting the nodes represent the direction of the heat threat transmission and carry a weight for the transmission risk level. The geometric center of mass refers to the geometric center of mass, which includes the center of mass X coordinate, the center of mass Y coordinate, and the center of mass Z coordinate. The X coordinate is obtained by taking the arithmetic mean of the X component of all heat source coordinates in the coordinate set, the Y coordinate is obtained by taking the arithmetic mean of the Y component of all heat source coordinates in the coordinate set, and the Z coordinate is obtained by taking the arithmetic mean of the Z component of all heat source coordinates in the coordinate set. The current pointing angle refers to the spatial direction of the optical boresight of the smart pod. The target pointing angle refers to the yaw and pitch angles that need to be corrected. The angular velocity component is the velocity projection of the difference between the target yaw angle and the current pointing angle in the yaw and pitch directions, constrained by the gimbal inertial parameters. Inertial parameters refer to the motion constraints of the magnetohydrodynamic gimbal (maximum angular velocity ±100° / s, maximum angular acceleration ±500° / s²).

[0144] In the embodiment of the present application, first, in step 1061, the spatial three-dimensional coordinates of all heat sources are extracted from each node in the threat propagation path, such as the motor module position (120, 80, 100) and the lens driver position (135, 95, 105), to form a complete coordinate data set; then, the geometric center of mass of the set is calculated, wherein an arithmetic average operation is independently performed on the X-axis coordinate, Y-axis coordinate, and Z-axis coordinate in the pod three-dimensional coordinate system, that is, the sum of the X-coordinate values ​​of all heat sources is divided by the number of heat sources to obtain the center of mass X component, and the center of mass Y component and Z component are calculated similarly; finally, the center of mass coordinate is output.

[0145] Secondly, the three-dimensional motion data of the intelligent pod body is integrated and collected through step 1062, wherein the magnetometer detects the geomagnetic direction to calculate the absolute horizontal yaw angle in the range of 0°~360°, and the three-axis accelerometer measures the gravity direction to calculate the vertical pitch angle in the range of -90°~90°; secondly, the gyroscope provides angle change rate data, and the dynamic motion error is compensated by the quaternion solution algorithm; finally, the three sensor data are fused in the data processing unit, and the Kalman filter technology is used to eliminate the instantaneous jitter error, and the smooth and accurate current pointing angle is output, for example, the horizontal yaw angle is accurate to ±0.1°, and the vertical pitch angle is ±0.05°.

[0146] Next, in step 1063, the geometric center of mass coordinates calculated in step 1061 are converted to the pod body coordinate system by applying a rigid body transformation matrix to translate the center of mass coordinate origin to the gimbal rotation center according to the pod installation attitude parameters, such as the horizontal tilt angle and the height offset; then, the horizontal projection deviation between the center of mass point and the Z axis is measured in the XOY plane of the pod coordinate system as the yaw correction reference, and the vertical projection deviation is measured in the YOZ plane as the pitch correction reference; then, a geometric projection relationship is established based on the pinhole imaging model in which the horizontal yaw angle correction is proportional to the inverse tangent ratio of the X-direction deviation distance and the focal length, and the vertical pitch angle correction is proportional to the inverse tangent ratio of the Y-direction deviation distance and the focal length, and the spatial relative relationship between the converted center of mass point and the center of the pod optical field of view is calculated; finally, the target pointing angle is output as the original pointing angle + angle correction.

[0147] Finally, step 1064 calculates the difference between the target pointing angle and the current pointing angle, decomposing the horizontal yaw error and the vertical pitch error into horizontal angular velocity components and pitch angular velocity components, respectively. Next, the inertial parameters of the magnetic fluid damping gimbal are read, including the maximum angular velocity limit (±100° / s) and the angular acceleration constraint (±500° / s²). Kinematic constraints are applied to the horizontal angular velocity components and the pitch angular velocity components. When the horizontal angular velocity component is greater than 100° / s, the velocity is limited, and the acceleration value is constrained to avoid inertial impact. A continuous angle change sequence is then generated based on equal divisions of the time axis, i.e., one control point every 20ms. Each point smoothly transitions through three stages: uniform acceleration, uniform speed, and uniform deceleration. While ensuring that neither the speed nor the acceleration exceeds the gimbal hardware limits, a discrete trajectory point sequence that satisfies the motion constraints is output. Finally, the intelligent pod is driven to execute the sequence, for example, starting point 0ms → time Angle A → Time Angle B→...→end point, so that the center of the optical field of view is continuously and stably aligned with the geometric center of mass coordinates within the set time.

[0148] For example, when a model A drone is performing an inspection at site B, the threat path is [motor (120, 80, 100) → camera (135, 95, 105)].

[0149] Specifically, the geometric center of mass is first calculated, and the result is (125.2, 90.5, 102.3); the pod is currently pointing at 45.2 and 12.8°; after coordinate transformation calculation, 0.72° and 0.44° need to be corrected; the gimbal is controlled to rotate smoothly at a horizontal angular velocity of 7.2° / s and a pitch angular velocity of 4.4° / s within 200ms, so that the center of the optical field of view is precisely aligned with the center of mass of the thermal threat.

[0150] This method achieves dynamic coupling of the threat path spatial characteristics and pod control, improves alignment accuracy through heat source energy-weighted center of mass positioning, and ensures stable tracking of the target by the line of sight using motion control constraints. Ultimately, it achieves continuous spatial synchronization between the optical field of view and the threat core area, significantly enhancing the high-definition monitoring capability of the heat conduction process.

[0151] The following is a complete example of steps 101 to 106:

[0152] During a nighttime urban security mission, a patrol drone equipped with an intelligent pod detected a swarm of three intruding drones in Area A (coordinates X:102.3, Y:37.8). First, the pod's infrared imaging module (384×288 resolution) captured multiple frames of thermal radiation images at 0.1 seconds per frame. Using optical flow tracking and NMF decomposition, three independent heat sources (heat sources 1-3) were decoupled. Heat source 1 exhibited a centroid displacement of ≤2 pixels and a directional variance of <15° over five consecutive frames. This generated a dynamic temperature field map showing a temperature gradient of 1.2°C / pixel within 2 seconds, from 78°C to 85°C.

[0153] Secondly, based on the temperature gradient vector, that is, the main direction of heat source 1 is 45° / amplitude 15℃ / cm, the spatial correlation is analyzed using a graph neural network, and the output correlation weight of heat source 1→2 is 0.82. Because the gradient direction is continuously pointing and the distance is <15cm, the significant thermal impact of the main motor (heat source 1) on the battery compartment (heat source 2) is identified.

[0154] Next, a spatial relationship model was constructed based on an association weight of 0.82. The connection weight was converted into a torque generation factor of 0.92. The 28Hz wind disturbance point torque amplitude was calculated to be 0.15 Nm / phase 48°, resulting in a synthesized damping torque sequence with a spectral matching degree of >97%. By comparing the original parameters of the magnetic fluid gimbal, the filter parameters were modified to a cutoff frequency of 26.1-36.1Hz and a gain factor of 1.23, reducing the gimbal jitter from ±1.2° to ±0.4°. Step 104: The gradient vector of heat source 1 was mapped to a 45° directional energy path, which intersected with the 120° path of heat source 2 at (0.22, 0.31). The initial interference intensity of the elliptical overlap region was calculated to be 57.6°C·cm². The modified filter parameters (cutoff frequency 26.1Hz / gain 1.23) were applied to retain 80% of the high-frequency components (intersection angle 75°>60°), outputting a calibrated interference intensity of 49°·cm². Step 105: Extract the thermal diffusion characteristics of heat source 1 (direction change sequence [+3°, +5°, -2°...], intensity fluctuation standard deviation 4.2°C / cm, thermal diffusion energy value 185.3 units). Calculate the characteristic similarity of heat source 1 to 2, which is 0.878 (trajectory matching 0.85 + fluctuation consistency 0.92). Multiply the characteristic similarity by the calibration interference intensity to obtain a propagation correlation weight of 43.0. Generate the threat propagation path [Heat source 1 → Heat source 2] with Heat source 1, the source with the highest energy, as the starting point (weight 21.7 for Heat source 2 → 3 < threshold 30).

[0155] Finally, the geometric centroid of the threat path (0.125, 0.225, 0.325) is calculated, and the target correction (yaw +0.72° / pitch +0.44°) is solved based on the pod's current pointing angle (yaw 42.5° / pitch 10.1°). A 200ms smooth deflection sequence (horizontal angular velocity 7.2° / s, pitch angular velocity 4.4° / s) is generated through kinematic constraints (angular velocity limit ±100° / s). The center of the optical field of view is driven to precisely align with the centroid (offset error <0.05°), achieving continuous tracking of the thermal threat transmission path.

[0156] Figure 2 The present invention provides a schematic diagram of a smart pod control system for a drone, as shown in FIG. Figure 2 As shown, the system includes:

[0157] The first generation module 21 is configured to obtain multiple frames of thermal radiation images of the UAV through the infrared imaging module in the intelligent pod, perform heat source decoupling processing on the multiple frames of thermal radiation images, and generate a dynamic temperature field map of each heat source;

[0158] An analysis module 22 is configured to analyze the spatial correlation between heat sources using a graph neural network based on the temperature gradient vectors of the dynamic temperature field maps of all the heat sources;

[0159] The correction module 23 is used to generate a damping torque that matches the wind disturbance spectrum based on the spatial correlation between the heat sources, and to correct the filtering parameters of the magnetic fluid damping pan / tilt platform in the intelligent pod based on the frequency response characteristics of the damping torque;

[0160] a calibration module 24 for mapping the temperature gradient vector to a corresponding energy transfer path and calibrating the interference strength between two adjacent energy transfer paths using the filtering parameters;

[0161] A second generating module 25 is configured to extract a heat diffusion feature from the temperature gradient vector and generate a threat propagation path according to the calibrated interference intensity and the heat diffusion feature;

[0162] The planning module 26 is used to dynamically plan the deflection angle of the smart pod according to the threat propagation path so that the center of the optical field of view is spatially synchronized with the geometric center of mass of the threat propagation path.

[0163] Figure 2 The intelligent pod control system for UAV can perform Figure 1 The implementation principles and technical effects of the intelligent pod control method for a drone described in the illustrated embodiment will not be elaborated on here. The specific manner in which each module and unit performs operations in the intelligent pod control system for a drone in the above-mentioned embodiment has been described in detail in the embodiments of the method and will not be elaborated on here.

[0164] In one possible design, Figure 2 The intelligent pod control system of the drone of the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0165] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0166] The processing component 32 is used for the above Figure 1 The intelligent pod control method of the drone in the embodiment.

[0167] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above method.

[0168] The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile memory device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0169] Of course, a computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0170] The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc.

[0171] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0172] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0173] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 1 The intelligent pod control method of the UAV of the embodiment shown is shown.

[0174] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0175] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0176] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0177] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for controlling an intelligent pod of a UAV, characterized in that: include: The infrared imaging module in the intelligent pod acquires multiple frames of thermal radiation images of the drone, performs heat source decoupling processing on the multiple frames of thermal radiation images, and generates a dynamic temperature field map of each heat source; Based on the temperature gradient vectors of the dynamic temperature field maps of all the heat sources, a graph neural network is used to analyze the spatial correlation between the heat sources; Based on the spatial correlation between heat sources, a damping torque matching the wind disturbance spectrum is generated, and the filtering parameters of the magnetic fluid damping pan / tilt platform in the intelligent pod are modified according to the frequency response characteristics of the damping torque; Mapping the temperature gradient vector to a corresponding energy transfer path, and using the filtering parameter to calibrate the interference intensity between two adjacent energy transfer paths; extracting a heat diffusion feature from the temperature gradient vector, and generating a threat propagation path based on the calibrated interference intensity and the heat diffusion feature; According to the threat propagation path, the deflection angle of the smart pod is dynamically planned so that the center of the optical field of view is spatially synchronized with the geometric center of mass of the threat propagation path.

2. The method according to claim 1, characterized in that Extracting a heat diffusion feature from the temperature gradient vector, and generating a threat propagation path according to the calibrated interference intensity and the heat diffusion feature, including: The temperature gradient vectors in the dynamic temperature field map of each heat source are decomposed into time series, and multiple features of the temperature gradient direction at continuous time points, such as the change trajectory, fluctuation pattern, and cumulative change amount, are extracted to form a direction change sequence, an intensity fluctuation sequence, and a heat diffusion energy value. Specifically, the direction change sequence is formed based on the change trajectory of the temperature gradient direction at continuous time points; the intensity fluctuation sequence is formed based on the fluctuation pattern of the temperature gradient magnitude at continuous time points; and the heat diffusion energy value is formed based on the cumulative change amount of the temperature gradient vector within a preset time window. combining the direction change sequence, the intensity fluctuation sequence and the heat diffusion energy value into a heat diffusion characteristic vector for each heat source; Based on the calibrated interference intensity, an interference intensity table between heat sources is constructed, where each table entry stores the interference intensity value between a pair of heat sources, and the characteristic similarity of the heat diffusion feature vectors of any two heat sources is calculated. The similarity is determined by fusing the trajectory repeatability and fluctuation consistency of the two heat sources. Multiplying the feature similarity with the interference intensity value of the corresponding heat source pair in the interference intensity table to obtain the propagation correlation weight between the heat sources; Taking the heat source with the largest heat diffusion energy value as the starting node, select the adjacent heat source with the highest propagation association weight with the current heat source from the heat sources that have not been traversed. Add the selected heat source to the path sequence and update it as the current heat source. Repeat the above steps until the propagation association weight is lower than the preset threshold or there are no optional adjacent heat sources. Output the path sequence as the threat propagation path.

3. The method according to claim 1, characterized in that Dynamically planning a deflection angle of the smart pod based on the threat propagation path so that the center of the optical field of view is spatially synchronized with the geometric center of mass of the threat propagation path, including: Obtaining the spatial coordinates of all heat sources in the threat propagation path to form a coordinate set, and calculating the geometric centroid of the coordinate set, wherein the geometric centroid includes the centroid coordinates of multiple components; Get the current pointing angle of the smart pod, where the current pointing angle includes the horizontal yaw angle and the vertical pitch angle; The center of mass coordinates are converted to the pod coordinate system through coordinate transformation. The horizontal yaw angle correction and vertical pitch angle correction are calculated based on the converted center of mass coordinates and the spatial relative position of the smart pod, and used as the target pointing angle; The difference between the current pointing angle and the target pointing angle is decomposed into angular velocity components. Kinematic constraints are applied to the angular velocity components based on the inertial parameters of the magnetofluid damping gimbal. A continuous angle change sequence that satisfies the constraints is output, and the smart pod is controlled to deflect according to the continuous angle change sequence so that the center of the optical field of view is aligned with the geometric center of mass coordinates.

4. The method according to claim 1, wherein The infrared imaging module in the intelligent pod acquires multiple frames of thermal radiation images of the drone, performs heat source decoupling processing on the multiple frames of thermal radiation images, and generates a dynamic temperature field map of each heat source, including: The infrared imaging module of the intelligent pod captures multiple frames of thermal radiation images of the drone at consecutive time points. Each frame of thermal radiation image contains the temperature values ​​of multiple pixel points. Tracking the heat source profile of each independent heat source in consecutive frames based on the radiation intensity distribution and spatial position stability of the multiple frames of thermal radiation images; Determining the radiation contribution of each independent heat source in the multiple frames of thermal radiation images using the tracked heat source profile; According to the radiation contribution, a unique identifier is assigned to each heat source, and its temperature distribution data in each frame of the image is extracted; the temperature distribution data of each heat source is integrated in time series to generate a dynamic temperature field map of the heat source.

5. The method according to claim 1, wherein Based on the temperature gradient vectors of the dynamic temperature field maps of all the heat sources, a graph neural network is used to analyze the spatial correlation between the heat sources, including: Extract the temperature gradient vector from the dynamic temperature field map of each heat source; Inputting the temperature gradient vector as a node feature into a graph neural network, wherein the graph neural network learns the dependency relationship between the node features; Through iterative calculation of the graph neural network, the association weights between nodes are output, and the association weights represent the degree of mutual influence of temperature changes between heat sources; A spatial correlation description between heat sources is generated according to the correlation weights.

6. The method according to claim 1, characterized in that Based on the spatial correlation between heat sources, a damping torque matching the wind disturbance spectrum is generated, and the filtering parameters of the magnetic fluid damping pan / tilt platform in the intelligent pod are corrected according to the frequency response characteristics of the damping torque, including: Based on the spatial correlation between the heat sources, a spatial relationship model is constructed, wherein the spatial relationship model represents the position of each heat source and the degree of mutual influence as a connection weight; Simulating the dynamic response of the wind disturbance spectrum at different frequencies through the spatial relationship model to convert the connection weights into moment generation factors, and calculating the moment amplitude and phase at multiple frequency points based on the moment generation factors; synthesizing a damping torque sequence according to the torque amplitude and phase, wherein a frequency spectrum characteristic of the damping torque sequence is consistent with a characteristic of the wind disturbance frequency spectrum; Comparing the amplitude response and phase response of the extracted frequency response characteristics of the damping torque sequence with current filtering parameters of the magnetic fluid damping pan / tilt platform in the intelligent pod to generate a parameter deviation; According to the parameter deviation, the cutoff frequency and the gain coefficient in the current filtering parameters are adjusted to generate modified filtering parameters.

7. The method according to claim 1, characterized in that Mapping the temperature gradient vector to a corresponding energy transfer path, and using the filtering parameter to calibrate the interference strength between two adjacent energy transfer paths, including: Taking each vector element in the temperature gradient vector as a starting point, performing path tracing along the vector direction to generate multiple energy transfer paths; calculating an overlapping area between two adjacent energy transfer paths, and quantifying the overlapping area as an initial interference intensity; The initial interference intensity is scaled and adjusted using the cutoff frequency and gain coefficient in the filtering parameters to determine the calibrated interference intensity, wherein the scaling and adjustment process includes: using the cutoff frequency to filter out high-frequency interference components, and using the gain coefficient to amplify or attenuate the amplitude of the initial interference intensity.

8. An intelligent pod control system for a drone, characterized in that: include: The first generation module is used to obtain multiple frames of thermal radiation images of the drone through the infrared imaging module in the intelligent pod, perform heat source decoupling processing on the multiple frames of thermal radiation images, and generate a dynamic temperature field map of each heat source; An analysis module for analyzing the spatial correlation between heat sources using a graph neural network based on the temperature gradient vectors of the dynamic temperature field maps of all the heat sources; A correction module is used to generate a damping torque that matches the wind disturbance spectrum based on the spatial correlation between heat sources, and to correct the filtering parameters of the magnetic fluid damping pan / tilt platform in the intelligent pod based on the frequency response characteristics of the damping torque; a calibration module, configured to map the temperature gradient vector to a corresponding energy transfer path, and to calibrate the interference intensity between two adjacent energy transfer paths using the filtering parameters; a second generating module, configured to extract a heat diffusion feature from the temperature gradient vector and generate a threat propagation path according to the calibrated interference intensity and the heat diffusion feature; A planning module is used to dynamically plan the deflection angle of the smart pod according to the threat propagation path so that the center of the optical field of view is spatially synchronized with the geometric center of mass of the threat propagation path.

9. A computing device, characterized in that It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement the intelligent pod control method of the drone as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, the intelligent pod control method of the drone according to any one of claims 1 to 7 is implemented.

Images