Salt pan unmanned aerial vehicle adaptive path dynamic planning method based on multi-source data fusion

Abstract

The invention relates to the technical field of artificial intelligence, in particular to a salt pan unmanned aerial vehicle adaptive path dynamic planning method based on multi-source data fusion, which comprises the following steps: dynamically regulating and controlling the sampling density of an unmanned aerial vehicle inspection path based on meteorological data and an evaporation rate prediction model; taking the regulated and controlled sampling density as input, and performing optimization adjustment on unmanned aerial vehicle track point distribution in combination with change characteristics of a salt crystallization region in the multispectral image; taking the optimized track point distribution as input, and performing adaptive adjustment on the flight height of the unmanned aerial vehicle according to real-time wind speed and direction and ground surface humidity information; the adjusted flight height is used as input, the path re-planning priority is intelligently judged in combination with a difference fusion result of satellite remote sensing data and an on-site image, and the switching opportunity of a flight path and a monitoring mode is controlled according to night illumination intensity and a thermal imaging contrast threshold value; the problem of inconsistent monitoring precision in a traditional fixed sampling mode is effectively solved, and more accurate data support is provided for evaporation capacity measurement and calculation.

Description

TECHNICAL FIELD

[0001] The application relates to the field of artificial intelligence technology, and in particular to a salt field unmanned aerial vehicle adaptive path dynamic planning method based on multi-source data fusion. BACKGROUND

[0002] To achieve adaptive dynamic planning of the salt field unmanned aerial vehicle inspection path, a multi-source data fusion technology is used to integrate meteorological data, multispectral images, real-time wind speed and direction, ground humidity, satellite remote sensing and night thermal imaging and other information to intelligently control the inspection path of the unmanned aerial vehicle.

[0003] In actual application, in view of the problem of inconsistent dynamic monitoring precision of salt field water evaporation, the sampling density is dynamically adjusted by combining meteorological data and evaporation rate prediction model to improve the monitoring precision; in order to avoid the decrease of inspection efficiency caused by misjudgment of salt concentration distribution, the system analyzes the change characteristics of the salt crystal region in the multispectral image to optimize the distribution of the flight path points; when facing a complex airflow environment, the flight height is adjusted according to the real-time wind speed and direction and the ground humidity to ensure the stability of data acquisition; considering that the path redundancy caused by the lag of large-scale salt field ground information update, the priority of path re-planning is adjusted by fusing the difference between satellite remote sensing and real image; in addition, in view of the target recognition misjudgment problem caused by the difference between day and night salt pool characteristics, the system intelligently switches the flight path and the monitoring mode according to the night lighting intensity and the thermal imaging contrast threshold. Through a series of measures, the overall accuracy and inspection efficiency of the salt field monitoring are comprehensively improved. SUMMARY

[0004] To solve the problems in the background art, the application provides a salt field unmanned aerial vehicle adaptive path dynamic planning method based on multi-source data fusion.

[0005] The application provides a salt field unmanned aerial vehicle adaptive path dynamic planning method based on multi-source data fusion, which adopts the following technical scheme:

[0006] The salt field unmanned aerial vehicle adaptive path dynamic planning method based on multi-source data fusion comprises the following steps:

[0007] Step 1: dynamically regulating the sampling density of the unmanned aerial vehicle inspection path based on meteorological data and an evaporation rate prediction model;

[0008] Step 2: optimizing and adjusting the distribution of the unmanned aerial vehicle flight path points by combining the change characteristics of the salt crystal region in the multispectral image based on the regulated sampling density as input;

[0009] Step 3: adaptively adjusting the flight height of the unmanned aerial vehicle according to real-time wind speed and direction and ground humidity information by taking the optimized flight path point distribution as input;

[0010] Step four, taking the adjusted flight height as input, intelligently determining the path re-planning priority in combination with the satellite remote sensing data and the difference fusion result of the field image, and controlling the switching timing of the flight path and the monitoring mode according to the night lighting intensity and the thermal imaging contrast threshold.

[0011] Preferably, the dynamic regulation of the sampling density of the unmanned aerial vehicle inspection path based on the meteorological data and the evaporation rate prediction model further comprises:

[0012] Obtaining a real-time meteorological data set P weather , including temperature T, relative humidity RH, and wind speed V;

[0013] Constructing an evaporation rate prediction model , wherein k is an empirical coefficient, T0 is a reference temperature, and n is an empirical power value;

[0014] Calculating the current regional evaporation rate deviation ΔE = |E real -E predicted |; wherein E real is the current evaporation rate, and E predicted is the historical reference value;

[0015] Adjusting the sampling density D according to the size of ΔE by the formula D = D base + α· ΔE, wherein D base is the basic sampling density, and α is the sampling density adjustment coefficient.

[0016] Preferably, the optimized adjustment of the unmanned aerial vehicle flight path point distribution taking the regulated sampling density as input in combination with the change characteristics of the salt crystallization region in the multispectral image further comprises:

[0017] Extracting the edge change amount ΔA of the salt crystallization region in the multispectral image;

[0018] Calculating the salt layer distribution map G using the gray difference method, defined as G(i,j) = |R(i,j) - R ref (i,j)|, wherein R(i,j) is the current band intensity, and R ref is the historical reference value;

[0019] Using a sliding window statistical method to perform gradient smoothing on the G value to generate a change characteristic map C;

[0020] According to the C value density distribution P C , dynamically adjusting the flight path point height H according to the formula H = max(H range · P C , H min ), wherein H range is the height range, and H min is the minimum limit value.

[0021] Preferably, the step of using the optimized waypoint distribution as input and adaptively adjusting the UAV flight altitude based on real-time wind speed, wind direction, and surface humidity information further includes:

[0022] Collect current wind speed V wind With wind direction θ;

[0023] Obtain the surface moisture index HI, and calculate it using NDVI data and the soil moisture content formula: HI = (NDVI) / (NDVI) max -NDVI current )·W soil ;

[0024] Using flight stability function To determine whether the flight altitude H needs to be increased / decreased;

[0025] Determine the height adjustment direction based on the following conditions: If F stability >F threshold If F, then increase H; stability <F threshold If β is used, then H is lowered (β is a stabilizing factor).

[0026] Preferably, the step of using the adjusted flight altitude as input and combining the results of the fusion of satellite remote sensing data and on-site images to intelligently determine the priority of path replanning further includes:

[0027] Calculate real-shot image I mixed With satellite remote sensing image I sat Mean square error MRE = (I mixed -I sat ) 2 ;

[0028] Extracting surface texture similarity index

[0029] According to MRE and T similarity Design priority determination formula: P priority =w1·MRE+w2·(1-T) similarity ), where w1 and w2 are weighting coefficients;

[0030] If P priority Greater than threshold T prio This triggers a path replanning instruction.

[0031] Preferably, the step of intelligently determining the priority of path replanning based on the fusion results of satellite remote sensing data and on-site images further includes:

[0032] Acquire nighttime thermal imaging images (Thermalimg) and identify the surface temperature difference ΔT of the target object;

[0033] The intensity of environmental interference is determined based on ΔT, and the thermal imaging contrast coefficient is calculated.

[0034] Combined with nighttime lighting intensity I light Construct the switching judgment function S switch =a·K+b·I light , where a is the weight of thermal imaging influence and b is the weight of illumination influence;

[0035] If S switch Less than threshold T switch If the condition is met, the system will enter high-precision monitoring mode; otherwise, the system will remain in regular monitoring mode.

[0036] Further adjustments to the timing of drone trajectory and monitoring mode switching based on nighttime lighting intensity and thermal imaging contrast threshold include:

[0037] Calculate the average light intensity within the target area

[0038] Extracting the contrast of a thermal imaging target δ = max(T) pixel )-min(T pixel );

[0039] The switching signal is determined based on the following function: Switch signal =(I avg threshold )∧(δ>δ threshold );

[0040] If Switch signal If true, the mode switching logic is triggered, and the target features are saved for subsequent backtracking analysis.

[0041] Preferably, the method of adjusting the timing of switching between UAV flight paths and monitoring modes based on nighttime lighting intensity and thermal imaging contrast threshold further includes:

[0042] Establish the nighttime illumination variation curve ΔI=I new -I old , where I new and I old These are the illuminance values ​​at two consecutive shooting times;

[0043] Extracting the rate of temperature change of thermal imaging targets Defined as the target dynamic factor;

[0044] Introducing a decision function:

[0045] If M decision ≥M threshold ​If the execution mode is switched, the target recognition database will be updated synchronously.

[0046] Preferably, the method according to claim 6 further refines the track mode switching strategy by including:

[0047] Establish an evaluation system for trajectory tracking models, taking into account path repetition rate. Target detection accuracy A acc ;

[0048] Constructing the switching cost function: Cost switch =w1·R repeated +w2·(1-A acc );

[0049] When Cost switch ≥Threshold cost At this time, the high-sensitivity monitoring mode will be forcibly activated;

[0050] After a successful switch, the feature vector of the current scene is recorded for subsequent automatic recognition of similar scenes and pattern matching.

[0051] Preferably, the step of adjusting the sampling density of the UAV inspection path based on meteorological data and an evaporation rate prediction model further includes:

[0052] Establish an evolution map of regional evaporation phenomena. trend ={ΔE1,ΔE2,...,ΔE n}, where ΔE i This represents the actual evaporation change over time period i;

[0053] Using the Fourier transform method to analyze E trend Frequency domain analysis was performed to obtain the frequency distribution F. freq ;

[0054] Calculate the spectral energy ratio As a criterion for climate anomalies;

[0055] Adjust the sampling frequency according to the following rules: If E ratio If E > α, then increase the number of sampling points. ratio If the value is less than β, then reduce the sampling density (α and β are threshold parameters).

[0056] Preferably, the method according to claim 8 further includes the following adjustment of the sampling frequency:

[0057] A dynamic sampling density control module is introduced to set the reference density D. initial ;

[0058] A density prediction model D is constructed using a neural network algorithm.predict =f(ΔE,F) freq E ratio ), where f is the pre-trained model function;

[0059] If the prediction result is D predict Above the upper limit D upper Perform a declassification operation; if the value is below the lower limit D lower Perform encryption operations;

[0060] All adjustments must ensure system stability to prevent the trajectory from deviating from the expected path due to frequent changes.

[0061] In summary, this application includes at least one of the following beneficial technical effects:

[0062] 1. The adaptive path dynamic planning method for salt field UAVs based on multi-source data fusion dynamically adjusts the sampling density through meteorological data and evaporation rate prediction models, making the UAV inspection sampling more in line with the real-time change pattern of evaporation on the salt field surface. This effectively solves the problem of inconsistent monitoring accuracy under the traditional fixed sampling mode and provides more accurate data support for evaporation measurement.

[0063] 2. This adaptive path dynamic planning method for salt field UAVs based on multi-source data fusion optimizes the distribution of track points by combining the changing characteristics of salt crystallization areas in multispectral images, reducing invalid inspection paths caused by misjudgment of salt concentration distribution. While ensuring monitoring coverage of key areas, it significantly improves inspection efficiency and reduces manpower and time costs.

[0064] 3. The adaptive path dynamic planning method for salt field UAVs based on multi-source data fusion adaptively adjusts the flight altitude according to real-time wind speed, wind direction and surface humidity, so that the UAV can maintain a stable flight attitude in complex airflow environment at low altitude, reduce the impact of airflow interference on sensors, and ensure the continuity and reliability of multi-source data (such as spectrum and image) acquisition.

[0065] 4. The adaptive path dynamic planning method for salt field UAVs based on multi-source data fusion can intelligently determine the priority of path replanning by fusing the differential results of satellite remote sensing and field images. It can quickly respond to the dynamic updates of salt field surface information (such as crystallization zone expansion, water level changes, etc.), avoid repeated inspections or missed inspections due to information lag, and optimize path resource allocation.

[0066] 5. The adaptive path dynamic planning method for salt field UAVs based on multi-source data fusion controls the switching of flight path and monitoring mode based on nighttime lighting intensity and thermal imaging contrast threshold, adapts to the differences in the characterization of salt ponds during the day and night (such as changes in temperature distribution and reflectivity), improves the accuracy of target identification at night, reduces misjudgments caused by ambient light interference, and ensures all-weather monitoring results.

[0067] 6. This adaptive path dynamic planning method for salt field drones based on multi-source data fusion integrates multi-dimensional data such as meteorological, multispectral, satellite remote sensing, and real-time environmental data to construct a full-process adaptive path planning mechanism. This promotes the transformation of salt field monitoring from "experience-driven" to "data-driven" and provides a scientific basis for salt field production management (such as brine production control and crystallization zone management). Attached Figure Description

[0068] Figure 1 This is a flowchart of an adaptive path dynamic planning method for UAVs in Yantian County based on multi-source data fusion. Detailed Implementation

[0069] The embodiments of this application are described in detail below, and examples of the embodiments are shown in the accompanying drawings.

[0070] In the description of this specification, the references to "certain embodiments," "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples" indicate that a specific feature, structure, material, or characteristic described in connection with an embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0071] This application discloses an adaptive path dynamic planning method for UAVs in salt fields based on multi-source data fusion, with reference to the appendix. Figure 1 This paper describes an adaptive path dynamic planning method for UAVs in salt fields based on multi-source data fusion, including:

[0072] Step 1: Dynamically adjust the sampling density of the UAV inspection path based on meteorological data and evaporation rate prediction model;

[0073] Step 2: Using the adjusted sampling density as input, and combining it with the variation characteristics of salt crystallization regions in the multispectral image, optimize and adjust the distribution of UAV track points;

[0074] Step 3: Using the optimized waypoint distribution as input, the drone's flight altitude is adaptively adjusted based on real-time wind speed, wind direction, and surface humidity information;

[0075] Step 4: Using the adjusted flight altitude as input, intelligently determine the priority of path replanning by combining the results of satellite remote sensing data and on-site image difference fusion, and control the timing of switching between the flight path and monitoring mode based on the nighttime lighting intensity and thermal imaging contrast threshold.

[0076] The dynamic adjustment of sampling density along the UAV inspection path based on meteorological data and evaporation rate prediction models further includes:

[0077] Obtain real-time weather dataset P weather This includes temperature T, relative humidity RH, and wind speed V;

[0078] Constructing an evaporation rate prediction model Where k is an empirical coefficient, T0 is the reference temperature, and n is an empirical power value;

[0079] Calculate the current regional evaporation rate deviation ΔE = |E real -E predicted |;

[0080] According to the formula D = D based on the magnitude of ΔE. base +α·ΔE adjusts the sampling density D, where D base The base sampling density is α, which is the sampling density adjustment coefficient.

[0081] Using the adjusted sampling density as input, and combining it with the variation characteristics of salt crystallization regions in multispectral images, the distribution of UAV track points is further optimized and adjusted, including:

[0082] Extract the edge variation ΔA of salt crystallization regions from multispectral images;

[0083] The salt layer distribution map G is calculated using the gray-level difference method, and G(i,j) = |R(i,j) - R ref (i,j)|, where R(i,j) is the current band intensity, R ref Historical reference value;

[0084] A sliding window statistical method is used to smooth the gradient of the G value, generating a change feature map C.

[0085] According to the C-value density distribution P C According to the formula H = max(H range ·P C H min Dynamically adjust the waypoint altitude H, where H range It is the height range, H min It is the minimum limit value.

[0086] Using the optimized waypoint distribution as input, and adaptively adjusting the UAV's flight altitude based on real-time wind speed, direction, and surface humidity information, further includes:

[0087] Collect current wind speed V wind With wind direction θ;

[0088] Obtain the surface moisture index HI, and calculate it using NDVI data and the soil moisture content formula: HI = (NDVI) / (NDVI) max -NDVI current )·W soil ;

[0089] Using flight stability function To determine whether the flight altitude H needs to be increased / decreased;

[0090] Determine the height adjustment direction based on the following conditions: If F stability >F threshold If F, then increase H; stability <F threshold If β is used, then H is lowered (β is a stabilizing factor).

[0091] Using the adjusted flight altitude as input, and combining the results of fusion of satellite remote sensing data and on-site image differences, the intelligent determination of path replanning priority further includes:

[0092] Calculate real-shot image I mixed With satellite remote sensing image I sat Mean square error MRE = (I mixed -I sat ) 2 ;

[0093] Extracting surface texture similarity index

[0094] According to MRE and T similarity Design priority determination formula: P priority =w1·MRE+w2·(1-T) similarity ), where w1 and w2 are weighting coefficients;

[0095] If P priority Greater than threshold T prio This triggers a path replanning instruction.

[0096] The intelligent determination of path replanning priority based on the fusion results of satellite remote sensing data and field images further includes:

[0097] Acquire nighttime thermal imaging images (Thermalimg) and identify the surface temperature difference ΔT of the target object;

[0098] The intensity of environmental interference is determined based on ΔT, and the thermal imaging contrast coefficient is calculated.

[0099] Combined with nighttime lighting intensity I light Construct the switching judgment function S switch =a·K+b·I light, where a is the weight of thermal imaging influence and b is the weight of illumination influence;

[0100] If S switch Less than threshold T switch If the condition is met, the system will enter high-precision monitoring mode; otherwise, the system will remain in regular monitoring mode.

[0101] Further adjustments to the timing of drone trajectory and monitoring mode switching based on nighttime lighting intensity and thermal imaging contrast threshold include:

[0102] Calculate the average light intensity within the target area

[0103] Extracting the contrast of a thermal imaging target δ = max(T) pixel )-min(T pixel );

[0104] The switching signal is determined based on the following function: Switch signal =(I avg threshold )∧(δ>δ threshold );

[0105] If Switch signal If true, the mode switching logic is triggered, and the target features are saved for subsequent backtracking analysis.

[0106] Further details on how to regulate the timing of drone trajectory and monitoring mode switching based on nighttime lighting intensity and thermal imaging contrast threshold include:

[0107] Establish the nighttime illumination variation curve ΔI=I new -I old , where I new and I old These are the illuminance values ​​at two consecutive shooting times;

[0108] Extracting the rate of temperature change of thermal imaging targets Defined as the target dynamic factor;

[0109] Introducing a decision function:

[0110] If M decision ≥M threshold If the execution mode is switched, the target recognition database will be updated synchronously.

[0111] Establish an evaluation system for trajectory tracking models, taking into account path repetition rate. Target detection accuracy A acc ;

[0112] Constructing the switching cost function: Cost​switch =w1·R repeated +w2·(1-A acc );

[0113] When Cost switch ≥Threshold cost At this time, the high-sensitivity monitoring mode will be forcibly activated;

[0114] After a successful switch, the feature vector of the current scene is recorded for subsequent automatic recognition of similar scenes and pattern matching.

[0115] Further adjustments to the sampling density of the UAV inspection path based on meteorological data and evaporation rate prediction models include:

[0116] Establish an evolution map of regional evaporation phenomena. trend ={ΔE1,ΔE2,...,ΔE n}, where ΔE i This represents the actual evaporation change over time period i;

[0117] Using the Fourier transform method to analyze E trend Frequency domain analysis was performed to obtain the frequency distribution F. freq ;

[0118] Calculate the spectral energy ratio As a criterion for climate anomalies;

[0119] Adjust the sampling frequency according to the following rules: If E ratio If E > α, then increase the number of sampling points. ratio If the value is less than β, then reduce the sampling density (α and β are threshold parameters).

[0120] By introducing a dynamic sampling density control module, a reference density D is set. initial ;

[0121] A density prediction model D is constructed using a neural network algorithm. predict =f(ΔE,F) freq E ratio ), where f is the pre-trained model function;

[0122] If the prediction result is D predict Above the upper limit D upper Perform a declassification operation; if the value is below the lower limit D lower Perform encryption operations;

[0123] All adjustments must ensure system stability to prevent the trajectory from deviating from the expected path due to frequent changes.

[0124] The adaptive path dynamic planning method for UAVs in salt fields based on multi-source data fusion includes several key steps:

[0125] First, it dynamically adjusts the sampling density of the drone's inspection path based on meteorological data and an evaporation rate prediction model. Specifically, it uses meteorological data acquired from ground weather stations or satellite remote sensing, combined with a pre-set evaporation rate prediction model, to calculate the trend of evaporation rate changes within a specific area. Then, based on the model output, it increases the sampling density in areas with high evaporation rates and appropriately decreases it when evaporation rates are low. For example, if a sudden temperature rise on a certain morning causes increased evaporation in some salt field areas, the system will automatically increase the sampling frequency of the flight path in that area to ensure high-precision acquisition of evaporation surface data.

[0126] Next, the adjusted sampling density is used as input, and combined with the changing characteristics of salt crystallization regions in the multispectral images, the distribution of UAV flight path points is optimized and adjusted. During operation, the system analyzes the images captured by the multispectral imaging system to identify the morphology and spatial distribution pattern of salt crystals. The system then combines historical observation data and current image information to determine whether the area of ​​salt crystallization has changed significantly. If the crystallization range is found to have expanded or shrunk, the spacing and coverage angle of the flight path points are adjusted accordingly to ensure that key areas are effectively monitored. For example, when the multispectral image shows a significant thickening of the crystal layer in a certain area, the system will adjust the path trajectory, allowing the UAV to stay in this area longer and repeat the scan more frequently.

[0127] Subsequently, the optimized waypoint distribution is used as input, and the drone's flight altitude is adaptively adjusted based on real-time wind speed, direction, and surface humidity information. During operation, the drone's onboard sensors collect real-time local wind speed, direction, and surface humidity data, which is then transmitted to the control center. Combining this with a pre-set aerodynamic model and surface disturbance patterns, the optimal flight altitude of the drone is dynamically adjusted. When high wind speeds or surface humidity may lead to poor atmospheric stability, the system will increase the flight altitude to avoid airflow interference; conversely, it will decrease the altitude to improve data clarity. For example, in the event of sudden gusts of wind, the drone's altitude adjustment can maintain stable image data quality, ensuring the accuracy of subsequent data analysis.

[0128] Next, using the adjusted flight altitude as input, and combining the results of fusion between satellite remote sensing data and on-site images, the system intelligently determines the priority of path replanning and controls the timing of switching between flight paths and monitoring modes based on nighttime lighting intensity and thermal imaging contrast thresholds. During implementation, the system first fuses real images captured by the ground-based UAV with satellite remote sensing images, analyzing the locations of differences. Since these differences may be caused by environmental changes or abrupt changes in surface conditions, the system prioritizes path replanning for these areas. Simultaneously, at night, the system uses night vision cameras or thermal imaging equipment to detect lighting conditions and thermal contrast thresholds. If the contrast is too high, potentially affecting target identification, the system automatically switches to a more suitable monitoring mode or changes the cruise path to prevent misjudgments. For example, during nighttime patrols, localized temperature differences often lead to increased thermal imaging contrast; the system proactively switches to an enhanced camera mode under low light conditions to ensure accurate identification and recording of targets in the salt lake area.

[0129] This method, which integrates multiple sensing technologies and data analysis algorithms, effectively improves the accuracy and flexibility of salt field monitoring, enabling high-quality, efficient, and stable drone operation processes in complex and dynamic environments.

[0130] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for adaptive path dynamic planning of UAVs in salt fields based on multi-source data fusion, characterized in that, include: Step 1: Dynamically adjust the sampling density of the UAV inspection path based on meteorological data and evaporation rate prediction model; Step 2: Using the adjusted sampling density as input, and combining it with the variation characteristics of salt crystallization regions in the multispectral image, optimize and adjust the distribution of UAV track points; Step 3: Using the optimized waypoint distribution as input, the drone's flight altitude is adaptively adjusted based on real-time wind speed, wind direction, and surface humidity information; Step 4: Using the adjusted flight altitude as input, intelligently determine the priority of path replanning by combining the results of satellite remote sensing data and on-site image difference fusion, and control the timing of switching between the flight path and monitoring mode based on the nighttime lighting intensity and thermal imaging contrast threshold.

2. The adaptive path dynamic planning method for salt field UAVs based on multi-source data fusion according to claim 1, characterized in that, The dynamic adjustment of sampling density along the UAV inspection path based on meteorological data and an evaporation rate prediction model further includes: Obtain real-time weather dataset P weather This includes temperature T, relative humidity RH, and wind speed V; Constructing an evaporation rate prediction model Where k is an empirical coefficient, T0 is the reference temperature, and n is an empirical power value; Calculate the current regional evaporation rate deviation ΔE = |E real -E predicted |; According to the formula D = D based on the magnitude of ΔE. base +α·ΔE adjusts the sampling density D, where D base The base sampling density is α, which is the sampling density adjustment coefficient.

3. The adaptive path dynamic planning method for salt field UAVs based on multi-source data fusion according to claim 2, characterized in that, The step of using the adjusted sampling density as input and combining it with the variation characteristics of salt crystallization regions in multispectral images to optimize and adjust the distribution of UAV flight points further includes: Extract the edge variation ΔA of salt crystallization regions from multispectral images; The salt layer distribution map G is calculated using the gray-level difference method, and G(i,j) = |R(i,j) - R ref (i,j)|, where R(i,j) is the current band intensity, R ref Historical reference value; A sliding window statistical method is used to smooth the gradient of the G value, generating a change feature map C. According to the C-value density distribution P C According to the formula H = max(H range ·P C H min Dynamically adjust the waypoint altitude H, where H range It is the height range, H min It is the minimum limit value.

4. The adaptive path dynamic planning method for salt field UAVs based on multi-source data fusion according to claim 3, characterized in that, The step of using the optimized waypoint distribution as input and adaptively adjusting the UAV's flight altitude based on real-time wind speed, wind direction, and surface humidity information further includes: Collect current wind speed V wind With wind direction θ; Obtain the surface moisture index HI, and calculate it using NDVI data and the soil moisture content formula: HI = (NDVI) / (NDVI) max -NDVI current )·W soil ; Using flight stability function To determine whether the flight altitude H needs to be increased / decreased; Determine the height adjustment direction based on the following conditions: If F stability >F threshold If F, then increase H; stability <F threshold If β is used, then H is lowered (β is a stabilizing factor).

5. The adaptive path dynamic planning method for salt field UAVs based on multi-source data fusion according to claim 4, characterized in that, The step of using the adjusted flight altitude as input and combining the results of the fusion of satellite remote sensing data and on-site image differences to intelligently determine the priority of path replanning further includes: Calculate real-shot image I mixed With satellite remote sensing image I sat The mean square error MRE = (I mixed -I sat ) 2 ; Extracting surface texture similarity index According to MRE and T similarity Design priority determination formula: P priority =w1·MRE+w2·(1-T) similarity ), where w1 and w2 are weighting coefficients; If P priority Greater than threshold T prio This triggers a path replanning instruction.

6. The adaptive path dynamic planning method for salt field UAVs based on multi-source data fusion according to claim 5, characterized in that, The intelligent determination of path replanning priority based on the fusion results of satellite remote sensing data and on-site images further includes: Acquire nighttime thermal imaging images (thermal images) and identify the surface temperature difference ΔT of the target object; The intensity of environmental interference is determined based on ΔT, and the thermal imaging contrast coefficient is calculated. Combined with nighttime lighting intensity I light Construct the switching judgment function S switch =a·K+b·I light , where a is the weight of thermal imaging influence and b is the weight of illumination influence; If S switch Less than threshold T switch If the condition is met, the system will enter high-precision monitoring mode; otherwise, the system will remain in regular monitoring mode. Further adjustments to the timing of drone trajectory and monitoring mode switching based on nighttime lighting intensity and thermal imaging contrast threshold include: Calculate the average light intensity within the target area Extracting the contrast of a thermal imaging target δ = max(T) pixel )-min(T pixel ); The switching signal is determined based on the following function: Switch signal =(I avg threshold )∧(δ>δ threshold );​ If Switch signal If true, the mode switching logic is triggered, and the target features are saved for subsequent backtracking analysis.

7. The adaptive path dynamic planning method for salt field UAVs based on multi-source data fusion according to claim 6, characterized in that, The method for adjusting the timing of switching between UAV flight paths and monitoring modes based on nighttime lighting intensity and thermal imaging contrast threshold further includes: Establish the nighttime illumination variation curve ΔI=I new -I old , where I new and I old These are the illuminance values ​​at two consecutive shooting times; Extracting the rate of temperature change of thermal imaging targets Defined as the target dynamic factor; Introducing a decision function: If M decision ≥M threshold If the execution mode is switched, the target recognition database will be updated synchronously.

8. The adaptive path dynamic planning method for salt field UAVs based on multi-source data fusion according to claim 7, characterized in that, The method according to claim 6 further refines the trajectory mode switching strategy by including: Establish an evaluation system for trajectory tracking models, taking into account path repetition rate. Target detection accuracy A acc ; Constructing the switching cost function: Cost switch =w1·R repeated +w2·(1-A acc ); When Cost switch ≥Threshold cost At this time, the high-sensitivity monitoring mode will be forcibly activated; After a successful switch, the feature vector of the current scene is recorded for subsequent automatic recognition of similar scenes and pattern matching.

9. The adaptive path dynamic planning method for salt field UAVs based on multi-source data fusion according to claim 8, characterized in that, The adjustment of the sampling density of the UAV inspection path based on meteorological data and evaporation rate prediction model further includes: Establish an evolution map of regional evaporation phenomena. trend ={ΔE1,ΔE2,...,ΔE n }, where ΔE i This represents the actual evaporation change over time period i; Using the Fourier transform method to analyze E trend Frequency domain analysis was performed to obtain the frequency distribution F. freq ; Calculate the spectral energy ratio As a criterion for climate anomalies; Adjust the sampling frequency according to the following rules: If E ratio If E > α, then increase the number of sampling points. ratio If the value is less than β, then reduce the sampling density (α and β are threshold parameters).

10. The adaptive path dynamic planning method for salt field UAVs based on multi-source data fusion according to claim 9, characterized in that, The method according to claim 8 further includes the following regarding the control of the sampling frequency: A dynamic sampling density control module is introduced to set the reference density D. initial ; A density prediction model D is constructed using a neural network algorithm. predict =f(ΔE,F) freq E ratio ), where f is the pre-trained model function; If the prediction result is D predict Above the upper limit D upper Perform a declassification operation; if the value is below the lower limit D lower Perform encryption operations; All adjustments must ensure system stability to prevent the trajectory from deviating from the expected path due to frequent changes.

Images