Reconstruction method and system for ecological restoration zoning of national space based on remote sensing surveying and mapping

Abstract

This invention relates to the field of remote sensing mapping technology, specifically to a method and system for reconstructing ecological restoration zones in national land space based on remote sensing mapping. The method includes: acquiring real-time meteorological data and real-time electrical data from a mapping drone, calculating the dynamic resonance boundary frequency of the mapping drone; acquiring high-frequency acceleration data along the Z-axis, performing frequency domain analysis with the dynamic resonance boundary frequency as the upper limit of integration, and extracting the intensity of physical disturbances; calculating the dynamic process noise covariance based on the intensity of physical disturbances, and updating it to the process noise covariance matrix of the navigation system; extracting exterior orientation elements using the navigation system and constructing a digital elevation model (DEM); extracting landform indicators using the DEM, thereby delineating ecological restoration zones and generating reconstruction instructions. This invention effectively eliminates airflow disturbances and improves the accuracy of terrain reconstruction under extreme conditions.

Description

Technical Field

[0001] This invention relates to the field of remote sensing mapping technology, and specifically to a method and system for reconstructing ecological restoration zones for national land space based on remote sensing mapping. Background Technology

[0002] Land space ecological restoration projects typically include peak shaving and valley filling, soil and water conservation, and ecological zoning. In order to provide accurate data support for subsequent engineering construction, it is necessary to obtain high-precision digital elevation models. At present, the industry generally uses drones equipped with surveying equipment to conduct photogrammetry, and uses the onboard navigation system to calculate the exterior orientation elements at the moment of camera exposure, thereby stitching two-dimensional images into a three-dimensional model. For example, Chinese patent application with publication number CN119085606A discloses a dynamic remote sensing photogrammetry system, which uses drone images to extract feature points and stitch them together to construct terrain data.

[0003] However, in actual surveying operations, especially when facing terrain with large elevation differences such as deep ravines or steep slopes, strong micro-airflow disturbances are prone to occur. These disturbances can cause low-frequency vibrations in surveying drones, leading to data drift and misalignment of topographic maps. To address this issue, existing technologies typically set a fixed isolation frequency threshold in the filtering stage to filter out low-frequency interference caused by external airflow. For example, Chinese patent document CN110596425B discloses a method for noise cancellation of a drone MEMS accelerometer, which uses a low-pass filter with fixed parameters to filter the acceleration values ​​for noise caused by the mechanical vibration of the drone.

[0004] However, long-term engineering practice has revealed that this fixed-threshold filtering mechanism is ill-suited to dynamic operating conditions. During extended surveying operations, the decrease in air density caused by high altitude and the voltage decay of the power battery at the end of its operation both lead to a decrease in the response stiffness of the electronic control system, causing the mechanical resonance frequency of the surveying drone to drift downwards. Because existing technology uses a fixed threshold, when the mechanical resonance frequency of the surveying drone coincides with this threshold, the filtering system will confuse the data, misjudging normal vibrations as external airflow interference. This results in the removal of a large amount of accurate data, leading to spatial distortion and elevation errors in the resulting digital elevation model. In subsequent ecological restoration zoning and earthwork calculations, the generated reconstruction commands will produce three-dimensional coordinate deviations. When automated excavation or leveling equipment at the construction site receives commands with these deviations, it will be unable to operate according to the expected planned boundaries, resulting not only in resource waste but also the potential for secondary geological risks due to over-construction. Summary of the Invention

[0005] To address the problem that existing technologies cannot adaptively adjust to complex terrain, this invention provides a method and system for reconstructing national spatial ecological restoration zones based on remote sensing mapping.

[0006] In a first aspect, the present invention provides a method for reconstructing national spatial ecological restoration zones based on remote sensing mapping, employing the following technical solution:

[0007] Real-time meteorological data and real-time electrical data of the surveying UAV in the target area are acquired and preprocessed to obtain a synchronized data sequence. A dynamic resonance boundary inference model is constructed, and the synchronized data sequence is input into the dynamic resonance boundary inference model to calculate the dynamic resonance boundary frequency of the surveying UAV. The Z-axis high-frequency acceleration data of the surveying UAV is acquired, and frequency domain analysis is performed on the Z-axis high-frequency acceleration data with the dynamic resonance boundary frequency as the upper limit of integration to extract the physical disturbance intensity. Based on the physical disturbance intensity, the dynamic process noise covariance is calculated, and the dynamic process noise covariance is updated to the preset process noise covariance matrix of the navigation system. The updated navigation system is used to extract exterior orientation elements, and a digital elevation model is constructed based on the exterior orientation elements. The digital elevation model is used to extract landform indicators, thereby delineating ecological restoration zones and generating reconstruction instructions.

[0008] This invention calculates the dynamic resonant boundary frequency of the surveying drone by combining real-time meteorological data of the target area and real-time electrical data inside the drone. This changes the existing technology's fixed threshold filtering method and can effectively distinguish between external airflow disturbances and the mechanical vibrations of the surveying drone itself. Furthermore, this invention converts the disturbance power of the airflow into the process noise covariance of the navigation system, enabling the filter to self-adjust according to changes in the external environment, effectively preventing calculation divergence problems in complex terrain.

[0009] Furthermore, the real-time meteorological data includes the real-time atmospheric pressure and real-time absolute temperature of the target area, and the real-time electrical data includes the real-time terminal voltage of the power battery of the mapping drone.

[0010] Furthermore, the preprocessing includes: acquiring a second pulse hardware signal, using the rising edge of the second pulse hardware signal as an absolute zero timestamp, and using an interpolation algorithm to upsample the real-time meteorological data and the real-time electrical data, aligning them to a high-frequency timestamp consistent with the Z-axis high-frequency acceleration data, thereby obtaining the synchronization data sequence.

[0011] This invention addresses the issue of inconsistent data refresh rates across different sensor hardware by using the second pulse hardware signal as the absolute zero point and employing an interpolation algorithm to process the collected data. This aligns external environmental features with internal state features in the time dimension, eliminating timing errors caused by communication bus delays and providing accurate data for subsequent calculations.

[0012] Furthermore, the method for reconstructing zoning for ecological restoration of national land space based on remote sensing mapping also includes: obtaining the specific gas constant of the target area, and calculating the in-situ atmospheric density based on the real-time atmospheric pressure, the real-time absolute temperature, and the specific gas constant.

[0013] Furthermore, the dynamic resonance boundary frequency satisfies the following relationship:

[0014]

[0015] in, For indexing time; For a moment The dynamic resonance boundary frequency; For a moment The aerodynamic resonant frequency; For electromechanical mapping factor; The nominal full-charge voltage of the power battery of the surveying drone; For a moment The real-time terminal voltage.

[0016] This invention converts the energy consumption of the power battery of a surveying drone into the loss ratio of mechanical stiffness, so that the calculated dynamic resonance boundary frequency can reflect the mechanical degradation of the surveying drone under high altitude and low power conditions, thus avoiding the system misjudging the normal vibration of the surveying drone itself as external interference.

[0017] Furthermore, the extraction of physical disturbance intensity includes: obtaining acceleration power spectral density, and for each moment, taking 0 as the lower limit and the dynamic resonance boundary frequency at that moment as the upper limit, performing an integral operation on the acceleration power spectral density to obtain the physical disturbance intensity at that moment.

[0018] Furthermore, the noise covariance of the dynamic process satisfies the following relationship:

[0019]

[0020] in, For indexing time; For a moment The dynamic process noise covariance; The initial process noise covariance; This is the energy noise expansion factor; For a moment The intensity of the physical disturbance.

[0021] This invention converts the intensity of physical disturbance into an expansion ratio. When a mapping drone encounters complex situations such as strong wind shear, the system can adaptively adjust the process noise covariance in the Z-axis direction of the filter, ensuring the convergence and stability of the navigation system calculation.

[0022] Furthermore, constructing a digital elevation model based on the exterior orientation elements specifically includes: acquiring remote sensing images of the target area; performing ray intersection operations on the remote sensing images carrying the exterior orientation elements to obtain corresponding points in three-dimensional space; removing non-surface interference data from the corresponding points in three-dimensional space; and fitting the data to generate the digital elevation model.

[0023] Furthermore, the method for reconstructing ecological restoration zones of national land space based on remote sensing mapping also includes: performing gridding processing on the digital elevation model, extracting the absolute slope and water catchment curvature of each grid as the landform indicators; dividing the target area into the ecological restoration zones according to preset ecological stability rules and the landform indicators, and generating the reconstruction instructions with geographic coordinates.

[0024] This invention uses a gridded analysis of a digital elevation model to extract the absolute slope and runoff curvature of the land surface, dividing the originally single topographic map into different functional restoration blocks, reducing survey errors and enabling subsequent operations to conform to ecological design boundaries.

[0025] Secondly, this invention provides a land space ecological restoration zoning reconstruction system based on remote sensing mapping, employing the following technical solution:

[0026] The land space ecological restoration zoning reconstruction system based on remote sensing mapping includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned land space ecological restoration zoning reconstruction method based on remote sensing mapping is implemented.

[0027] By adopting the above technical solution, the above-mentioned method for reconstructing the ecological restoration zoning of national land space based on remote sensing mapping is generated into a computer program and stored in a container so that it can be loaded and executed by a processor. In this way, a terminal device can be made based on the memory and processor for convenient use.

[0028] The present invention has the following technical effects:

[0029] This invention addresses the limitations of existing technologies that rely on static filtering, which is ill-suited to the hardware degradation and environmental abrupt changes caused by long-term, high-altitude fluctuations. By comprehensively considering environmental meteorological characteristics and electromechanical operating conditions, the isolation boundary of the filter can be dynamically adjusted in real time, thus avoiding the invalid removal of data.

[0030] This invention transforms processed high-precision location information into ecological zoning instructions to guide excavation and leveling machinery in construction, effectively reducing the rate of rework and the risk of over-construction in the ecological restoration process, and providing reliable technical support for large-scale terrain restoration. Attached Figure Description

[0031] Figure 1 This is a flowchart of the method for reconstructing national spatial ecological restoration zones based on remote sensing mapping provided in an embodiment of the present invention;

[0032] Figure 2 This is a comparison diagram of the effects of the prior art provided in the embodiments of the present invention and the present invention. Detailed Implementation

[0033] This invention provides a method for reconstructing national spatial ecological restoration zones based on remote sensing mapping, referring to... Figure 1 This includes steps S1-S4:

[0034] S1: Data Acquisition and Preprocessing.

[0035] Specifically, real-time meteorological data and real-time electrical data of the target area are acquired. The second pulse hardware signal of the navigation system is used as the absolute zero-point timestamp. An interpolation algorithm is used to upsample the real-time meteorological data and real-time electrical data and align them to a high-frequency timestamp consistent with the Z-axis high-frequency acceleration data to obtain a synchronized data sequence.

[0036] To accurately assess the aerodynamic environment of the target area, real-time meteorological data is required, including real-time atmospheric pressure and real-time absolute temperature of the target area, as detailed below:

[0037] In order to reduce the interference of the downwash airflow generated when the rotor of the surveying UAV rotates at high speed, this embodiment uses a static pressure Pitot tube installed on the windward side of the UAV arm to collect the real-time atmospheric pressure of the target area.

[0038] Meanwhile, this embodiment uses an external temperature probe to collect the real-time absolute temperature of the target area, which is deployed in the shaded and ventilated area on the underside of the surveying drone. This installation method can effectively block the heat radiation generated by the high-voltage power motor and ESC module during long-term high-speed operation, ensuring that the collected real-time absolute temperature reflects the changes in the external climate.

[0039] Next, in order to assess the electrical health status and thrust maintenance capability of the mapping drone, it is necessary to obtain the real-time electrical data of the mapping drone. In this embodiment, the battery management module embedded in the power battery of the mapping drone is selected. The real-time terminal voltage in the underlying register of the battery management module is read in real time through the industrial control bus. The real-time terminal voltage can reflect the depth of power drop of the mapping drone during long-term operation.

[0040] In addition, in order to assess the vertical mechanical destructive force of strong wind shear on the surveying drone, it is necessary to obtain the three-dimensional mechanical response characteristics of the surveying drone. In this embodiment, a high-frequency inertial measurement module integrated into the navigation system of the surveying drone is selected. Using its internal MEMS accelerometer, the high-frequency acceleration data of the surveying drone body on the Z-axis, that is, the Z-axis direction perpendicular to the ground plane, is output.

[0041] Due to differences in the underlying analog-to-digital conversion chips and communication baud rates between various external sensors and the internal bus, the refresh cycles of the acquired data output are inconsistent. For example, limited by bus bandwidth allocation, the sampling frequency for real-time atmospheric pressure and real-time terminal voltage is 50Hz, while the sampling frequency for Z-axis high-frequency acceleration data is 1000Hz. If these data are processed directly, control causality misalignment and error divergence will occur. Therefore, the acquired data needs to be aligned, as follows:

[0042] First, the rising edge of the second pulse hardware signal is obtained. In the hardware architecture of the mapping UAV, the second pulse hardware signal comes from the underlying satellite timing module. Since this signal is a pure physical level signal, it is not affected by any flight control system software thread blocking or communication scheduling, and has a high degree of time determinism. In this embodiment, the rising edge of the signal is set as the absolute zero point timestamp on the global time axis to reset the timing of the collected data.

[0043] After determining the absolute zero timestamp, this embodiment uses cubic spline interpolation algorithm to process sparse low-frequency data. By taking the actual sampling time of the low-frequency data as the interpolation node, an independent third-order polynomial is constructed for any time interval formed by two adjacent interpolation nodes. Considering the natural laws of smooth and gradual changes in weather and electrical states, such as the thermal inertia of temperature changes and the capacitive smoothing effect of battery voltage fluctuations, when solving the mathematical problem, it is constrained that not only is the function value equal to the actual sampled value at all interpolation nodes, but its first and second derivatives must also remain continuous. This constraint can effectively avoid sharp corners at the data connection points and prevent false high-frequency noise from appearing in subsequent calculations.

[0044] Next, using the 1000Hz time axis where the Z-axis high-frequency acceleration data is located as the query reference, i.e., a high-frequency timestamp is generated every 1ms, and these high-frequency timestamps are substituted into the third-order polynomial constructed above for solving; through this millisecond-level calculation based on time, the system upsamples the real-time atmospheric pressure, real-time absolute temperature, and real-time terminal voltage to obtain data with the same frequency as the Z-axis high-frequency acceleration data;

[0045] Finally, these scattered physical quantities from different sources and with varying rates are integrated into a set of synchronized data sequences aligned in the time dimension, providing clean data for subsequent operations and reducing the risk of filter divergence caused by the asynchronous timing of the basic data.

[0046] It should be noted that the hardware devices and algorithms selected for data acquisition and preprocessing mentioned above are merely preferred examples for achieving the technical objectives of this invention. In practical applications, the acquisition of real-time meteorological data is not limited to static pressure pitot tubes and single temperature probes; it can also be achieved by connecting an external integrated micro atmospheric data computer or using a distributed temperature and pressure sensor array. The acquisition of real-time terminal voltage can also be achieved by connecting independent high-precision voltage isolation transmitters in series. During preprocessing, the determination of the absolute zero-point timestamp is not limited to a second pulse hardware signal; it can also use a precision time synchronization protocol or a global synchronization frame of time-triggered Ethernet. At the same time, the processing of low-frequency data is not limited to cubic spline interpolation. In edge computing nodes with limited computing power, it can be equivalently replaced by linear interpolation, Lagrange interpolation, or dynamic prediction compensation based on Kalman state observers. Any method that uses a unified global clock reference to align meteorological data with electrical data is included within the scope of protection of this invention.

[0047] S2: Calculation of dynamic resonance boundary frequency.

[0048] Specifically, the specific gas constant of the target area is obtained, and the in-situ atmospheric density is calculated based on the real-time atmospheric pressure and real-time absolute temperature, combined with the specific gas constant. A dynamic resonance boundary inference model is constructed, and the synchronous data sequence is input into the dynamic resonance boundary inference model. Combined with the in-situ atmospheric density, the dynamic resonance boundary frequency of the mapping UAV is calculated.

[0049] Since the rotors of surveying drones generate upward aerodynamic thrust by displacing air downwards, the magnitude of this thrust depends on the mass of air displaced by the rotors, which is limited by the atmospheric density of the surrounding space. In areas with large elevation changes, such as the Loess Plateau or high mountain valleys, the dramatic changes in altitude lead to a significant decrease in atmospheric density. Therefore, it is necessary to perform dimensionality reduction and fusion on the collected data to extract the hydrodynamic index that determines thrust efficiency, namely, the in-situ atmospheric density, as follows:

[0050] According to the macroscopic ideal gas law Gas pressure With volume The product equals the number of moles of gas. Universal gas constant Thermodynamic temperature of gas The product; due to the number of moles of gas Equal to the total mass of the gas Divided by the average molar mass of the gas The ideal gas law can be equivalently replaced by: In thermodynamic engineering, the universal gas constant The ratio of the gaseous mass to the average molar mass is defined as the gaseous constant, which represents the work done per unit mass of gas to increase the temperature by 1 K.

[0051] By rearranging and simplifying the above ideal gas law, we obtain: ,in, For indexing time, For a moment In-situ atmospheric density, The specific gas constant for the target region is taken as 287.05 J / (kg·K) in this embodiment. For a moment Real-time atmospheric pressure. For a moment The real-time absolute temperature.

[0052] To obtain the fundamental mechanical excitation frequency for maintaining hovering operations of the surveying UAV under different atmospheric densities, this step constructs the aerodynamic resonance frequency based on the rotor momentum theory and static equilibrium law in aerodynamics, as detailed below:

[0053] According to the static equilibrium equation, the thrust of the mapping UAV when hovering is numerically equal to the product of the total mass and the acceleration due to gravity. At the same time, according to the thrust equation of rotor momentum theory, the magnitude of this thrust is equal to the product of the thrust coefficient, the in-situ atmospheric density, the rotor disk area, and the square of the rotor blade tip linear velocity. By equivalently combining the above static equilibrium equation and thrust equation, the rotational speed required for the mapping UAV to maintain its current equilibrium state can be obtained. According to the laws of rotational machinery vibration dynamics, multiplying this rotational speed by the number of propeller blades can be converted into the blade passage frequency of the mapping UAV in the hovering condition, which is defined as the aerodynamic resonance frequency. This frequency is the physical fundamental frequency source that causes the body resonance.

[0054] It should be noted that in the above deduction process, the core calculation variables fall into two categories: one is environmental variables that are limited by meteorological conditions and require real-time high-frequency sampling, and the other is static physical quantities that are limited by the mechanical structure of the surveying UAV. Within the ideal fluid dynamics framework, these static physical quantities can be regarded as known constants. However, in surveying operations, the rotor thrust coefficient and effective aerodynamic surface area are easily subjected to nonlinear distortion due to the influence of blade manufacturing tolerances and high-frequency aerodynamic deformation. Therefore, in this embodiment, they are folded as a whole in the algebraic structure, equivalently mapped and encapsulated into a macroscopic characteristic quantity, defined as an aerodynamic structure constant, and calibrated through a static thrust simulation experiment. The specific steps are as follows:

[0055] First, a static thrust test bench was built in a completely enclosed microclimate environment control cabin. The same type of motor and propeller as the actual carrier of the mapping UAV were fixed to the end of the cantilever of the test bench by a rigid metal flange. A high-precision strain gauge tensile sensor was installed in series between the motor and the cantilever, and the hardware data refresh rate of the tensile sensor was set to 1000Hz.

[0056] Subsequently, the microclimate environment control cabin was sealed, and the vacuum pump and temperature-controlled compressor inside the cabin were started to adjust the environment to the same high-altitude, low-pressure conditions as the target area where the mapping drone was located. After the airflow inside the cabin was completely still, the integer values ​​of the air pressure and temperature inside the cabin were read. Based on the aforementioned formula for calculating in-situ atmospheric density, the reference atmospheric density inside the cabin was calculated and denoted as . ;

[0057] Next, using a high-precision electronic scale, the total mass of the mapping UAV under full-load operation was measured. This mass was then multiplied by the gravitational acceleration of the target area to obtain the absolute static thrust required for the mapping UAV to maintain mechanical equilibrium. This static thrust was defined as the target mechanical threshold and denoted as [missing information]. ;

[0058] The electronic speed controller of the motor sends control commands with progressively increasing duty cycles to drive the propeller to accelerate its rotation. In this embodiment, an STM32 series microcontroller is used as the lower-level signal source. Its advanced hardware timer channel is used to generate a pulse width modulation signal with a base frequency of 400Hz. First, an initial drive command with a duty cycle of 10% is output to trigger the motor to overcome friction and enter idle operation. Then, the duty cycle of the output pulse width modulation signal is increased by 0.1% in a fixed increment over a 50ms time window. As the speed gradually increases, the tension value output by the high-precision strain gauge tension sensor is monitored in real time. When the tension value is equal to the target mechanical threshold, the control command of the current electronic speed controller is locked to keep the motor in the current state.

[0059] A laser tachometer pre-aligned with the propeller blades is invoked. This tachometer emits a laser beam and receives the reflected pulses from the blades, capturing the number of times the propeller cuts the laser beam per second. This number is then converted into the reciprocal of the mechanical vibration period to obtain the mechanical resonant frequency of the motor and propeller in the current state, denoted as . ;

[0060] Based on the aforementioned process for constructing aerodynamic resonance frequencies, the aerodynamic structural constants are calculated. , .

[0061] After obtaining the aerodynamic structure constant, it is multiplied by the overall mass and gravitational acceleration of the mapping UAV, and the in-situ atmospheric density is used as the denominator. The arithmetic square root of the fraction is then performed to obtain the real-time aerodynamic resonance frequency, denoted as . .

[0062] Because the mechanical resonance frequency of a surveying drone during operation is not a fixed value, but rather drifts bidirectionally with external weather conditions and internal power supply status; on the one hand, when the altitude increases and the air becomes thinner, in order to maintain the thrust required to counteract gravity, the motor and rotor of the surveying drone need to operate at a higher speed, which will raise the mechanical resonance frequency of the surveying drone body; on the other hand, as the operation continues, the available power of the surveying drone's power battery is constantly consumed and the terminal voltage decays, resulting in a decrease in the adjustment stiffness of the electronic control system when responding to external interference, which will lower the mechanical resonance frequency.

[0063] Based on the above patterns, a dynamic resonance boundary derivation model is constructed, and the dynamic resonance boundary frequency is calculated. The specific relationship is as follows:

[0064]

[0065] in, For indexing time; For a moment The dynamic resonant boundary frequency; For a moment The aerodynamic resonant frequency; The electromechanical mapping factor characterizes the rate of mechanical stiffness degradation caused by a unit voltage drop; The nominal full-charge voltage of the power battery for the surveying drone; For a moment Real-time terminal voltage, The item represents the percentage of loss caused by the attenuation of the supply voltage.

[0066] It should be noted that the electromechanical mapping factor It was calibrated through a dynamic voltage drop test, and the specific steps are as follows:

[0067] First, an electromechanical dynamometer test platform is established. A motor of the same model as the surveying drone is fixed on the test axis of the dynamometer. The electronic speed controller that controls the motor is connected to the data acquisition terminal of the dynamometer. At the power supply end, a high-precision programmable DC power supply is connected and its output voltage is set to the nominal full-charge voltage of the power battery of the surveying drone, which represents that the surveying drone is in an ideal power supply state with 100% power.

[0068] Next, a set of dynamic disturbance commands with standard sweep frequency characteristics are continuously sent to the electronic speed controller via the pulse width modulation bus. In this embodiment, an arbitrary waveform generator with a built-in direct digital frequency synthesis chip is selected. The standard sweep frequency characteristic is set to a linear sinusoidal frequency modulation wave signal. The starting frequency of the signal is set to 1Hz, the cutoff frequency is set to 50Hz, the single scan period is 10s, and the amplitude of the signal is set to ±5% of the basic hovering duty cycle to ensure that the dynamic countermeasure process of the motor when the mapping UAV encounters high-altitude gusts can be reproduced.

[0069] While sending the above instructions, the high-frequency torque sensor of the dynamometer records the torque waveform of the motor output in real time, extracts the timestamp of the peak, subtracts the timestamp of the peak in the linear sinusoidal frequency modulation signal, and obtains the torque response delay time. Since the current power supply is the nominal full voltage, the torque response delay time is defined as the ideal full stiffness.

[0070] Subsequently, the high-precision programmable DC power supply is controlled to reduce its output voltage to a preset test low voltage value, which is the safety threshold for the automatic return-to-home function of the mapping UAV when the battery is low. Under this test low voltage value, the same dynamic disturbance command is sent to the electronic speed controller again, the corresponding torque waveform is recorded, and the torque response delay time in the current state is calculated. Since the motor drive capability is limited due to the voltage drop, the torque response delay time in the current state is greater than the ideal full stiffness. The difference between the two is obtained, and the difference is divided by the ideal full stiffness to obtain the delay degradation percentage. Since the increase in response delay is physically equivalent to the change in control stiffness, the delay degradation percentage is defined as the stiffness degradation ratio.

[0071] Next, the difference between the nominal full-voltage and the tested low-voltage value is calculated. This difference is divided by the nominal full-voltage to obtain the voltage sag ratio. The stiffness degradation ratio is then divided by the voltage sag ratio to obtain the electromechanical mapping factor. .

[0072] S3: Dynamic correction of process noise covariance.

[0073] Specifically, using the dynamic resonance boundary frequency as the upper limit of integration, frequency domain analysis is performed on the Z-axis high-frequency acceleration data to extract the physical disturbance intensity; based on the physical disturbance intensity, the dynamic process noise covariance is calculated, and the dynamic process noise covariance is updated to the preset process noise covariance matrix of the navigation system for dynamic correction.

[0074] First, high-frequency acceleration data along the Z-axis, representing the strength of vertical vibration of the UAV body, is extracted from the synchronous data sequence. Then, a dynamically truncated data interval is constructed, defined as a sliding time window with a fixed length of 512 sampling points, corresponding to the motion state span of the past 0.512 seconds.

[0075] Subsequently, the Fast Fourier Transform algorithm is used to perform a basis transformation algebra operation from the time domain to the frequency domain on the data within the sliding time window. The complex body vibration that was originally in a single time dimension is decomposed into a series of sinusoidal waveforms with different frequencies and independent amplitudes, resulting in a frequency domain complex sequence that is strongly correlated with the frequency. Each complex element in this sequence represents the absolute vibration amplitude and phase information of the UAV body at a specific frequency point.

[0076] Since complex numbers contain an imaginary part, they cannot directly represent the actual strength of kinetic energy at the physical level. Therefore, the system extracts the absolute value of the complex amplitude corresponding to each frequency point in the frequency domain complex number sequence, and performs a square operation on each absolute value to convert the complex amplitude into an absolute value that characterizes the strength of vibration energy.

[0077] Next, the total sampling frequency, i.e., 1000Hz predefined in S1, is divided by the length of the sliding time window to obtain the frequency resolution; the absolute value representing the strength of energy obtained above is divided by the frequency resolution to obtain the acceleration power spectral density, which represents the density of vertical vibration energy borne by the UAV body within a unit Hertz frequency band at any frequency.

[0078] Because the high-frequency inertial measurement module of a mapping UAV experiences a wide range of physical vibration frequencies during operation, some higher-frequency vibration components, such as high-frequency white noise generated by high-speed motor operation and high-frequency aerodynamic noise generated by high-speed propeller cutting through the air, are naturally attenuated and filtered out by the structural stiffness of the frame during transmission. These high-frequency noises do not have a substantial impact on subsequent analysis. Conversely, what truly causes the underlying state estimator to diverge and cause elevation distortion are the low-frequency vibration components concentrated in the low-frequency band and located precisely in the boundary region. Based on this, this step constructs a frequency domain definite integral model with the dynamic resonance boundary frequency as the truncation threshold. Using the previously obtained acceleration power spectral density, the physical disturbance intensity is calculated. Specifically, with 0 as the initial lower limit and time... The dynamic resonance boundary frequency is taken as the upper cutoff frequency. By integrating the acceleration power spectral density, the time is obtained. The physical disturbance intensity is denoted as .

[0079] In traditional flight control systems for mapping UAVs, the process noise covariance of the underlying state estimator is typically hard-coded as an empirical constant. When a mapping UAV encounters severe resonance in a complex operating environment, its internal prediction model will experience trajectory deviation and distortion. If the process noise covariance remains unchanged, the underlying state estimator will continue to rely on the erroneous predicted trajectory, leading to divergent calculation results. Therefore, it is necessary to use the aforementioned physical disturbance intensity to calculate the dynamic process noise covariance and correct it in real time. The specific relationship is as follows:

[0080]

[0081] in, For indexing time; For a moment The dynamic process noise covariance; The initial process noise covariance represents the process noise covariance of a surveying UAV operating safely and stably in a windless and stationary state. This is the energy noise expansion factor; For a moment The intensity of physical disturbance.

[0082] It can be seen that by using the energy noise expansion factor, The term is transformed into a proportional expansion coefficient, which is then multiplied by the initial process noise covariance to obtain the process noise covariance adaptively adjusted based on the disturbance intensity.

[0083] After obtaining the dynamic process noise covariance, the system calls the pre-built navigation system inside the mapping UAV. In this embodiment, the satellite integrated navigation solution module is preferred. The system locates the process noise covariance matrix in its underlying filtering algorithm, removes the original constant elements on its diagonal that represent the variance of the Z-axis state prediction, and overwrites and replaces the obtained dynamic process noise covariance in real time. In the subsequent state prediction and measurement update process, based on the locally magnified matrix elements, the trust weight of the internal prediction model with severe body resonance is reduced, and the trust weight of external absolute position sensors that are not affected by the high-frequency vibration of the body, such as the RPT high-precision positioning antenna, is increased. Through this dynamic adaptive flow of underlying trust weights, the accuracy of spatial elevation and pose calculation of the mapping UAV under any working conditions is guaranteed.

[0084] It should be noted that the energy noise expansion factor was calibrated through a dynamic high-frequency vibration comparison experiment, the specific steps of which are as follows:

[0085] First, a high-frequency vibration test platform is built in a windless, static laboratory. The mapping UAV is fixed on the test platform of a six-degree-of-freedom electromagnetic vibration table, and a sub-millimeter precision optical motion capture system is set up around the test platform. In the initial stage of testing, the six-degree-of-freedom electromagnetic vibration table is kept stationary. The state estimator at the bottom layer of the mapping UAV is run, while the optical motion capture system is controlled to track and output a reference height program sequence generated by sensor white noise in real time. By comparing the calculation results of the state estimator with the reference height program sequence, the covariance value inside the system is repeatedly fine-tuned until the difference between the two approaches 0. In this embodiment, a difference of less than 10 is preferred. -4 The covariance value at this point is the initial process noise covariance.

[0086] A six-degree-of-freedom electromagnetic vibration table is started, and its output is controlled to fully cover the dynamic resonance boundary frequency of a broadband vibration. Under this vibration condition, the high-frequency inertial measurement module of the mapping UAV outputs a time-domain sequence containing vibration noise in real time. The same algorithm logic as in S2-S3 above is called to calculate the physical disturbance intensity, defined as the nodal disturbance energy, denoted as . ;

[0087] Next, using the initial process noise covariance as the starting reference value, 50 covariance test values ​​are continuously generated upwards in an arithmetic progression, increasing by 5% of the value. The raw data output in real time from the high-frequency inertial measurement module is extracted as a fixed input source. The covariance test values ​​are replaced one by one and input into the underlying state estimator for offline dead reckoning. For each covariance test value, the state estimator will recalculate and output a unique offline high-frequency procedure sequence.

[0088] For each offline high-precision sequence, extract the elevation data point under each discrete timestamp within it, subtract it from the elevation data point under the same timestamp in the benchmark high-precision sequence output by the optical motion capture system, take the absolute value of the subtraction result to obtain the single-point elevation error under a single timestamp, calculate the arithmetic mean of the single-point elevation errors under all timestamps of each offline high-precision sequence to obtain the mean absolute error.

[0089] By comparing the magnitudes of all mean absolute errors (MAEs) horizontally, the covariance test value with the smallest MAE is extracted. This value represents the optimal value that can suppress and converge the current vibration divergence to the greatest extent, and is defined as the optimal process noise covariance, denoted as . ;

[0090] Based on the aforementioned relationship of dynamic process noise covariance, the energy noise expansion factor is obtained by rearranging the terms. .

[0091] S4: Refactoring instruction generation.

[0092] Specifically, the updated navigation system is used to extract exterior orientation elements, a digital elevation model is constructed based on the exterior orientation elements, and landform indicators are extracted using the digital elevation model to delineate ecological restoration zones and generate reconstruction instructions.

[0093] During the operation of the surveying drone, the camera on it continuously captures remote sensing images of the target area. During each capture, the system extracts the camera's coordinates in three-dimensional space and the three-dimensional spatial attitude angles from the updated navigation system. The three-dimensional spatial attitude angles include roll angle, pitch angle, and yaw angle. The above coordinates and attitude angles are defined as the exterior orientation elements of the current remote sensing image.

[0094] Subsequently, the system extracts two adjacent remote sensing images with overlapping shooting areas, extracts the pixel coordinates pointing to the same physical feature point on the ground surface, and defines them as corresponding pixels. Based on the exterior orientation elements, the system reconstructs the spatial position and attitude of the mapping UAV when shooting these two remote sensing images in a three-dimensional coordinate system. Starting from the center point of the camera, the system projects two spatial rays through the corresponding pixels on the respective remote sensing images and projects them towards the ground surface. The system then solves for the intersection of these two spatial rays, and the three-dimensional coordinates of this intersection point are the location of the real feature point on the ground surface.

[0095] Repeat the above ray intersection operation for all corresponding pixels in these two remote sensing images to obtain a set of corresponding 3D spatial points covering the target area. This set contains all protrusions within the target area, including not only the actual bare ground but also non-surface disturbance data such as vegetation canopies and temporary structures. To accurately assess the terrain, these non-surface disturbance data need to be removed, as follows:

[0096] The entire set of corresponding points in three-dimensional space is divided into multiple local two-dimensional search blocks on a horizontal plane. In this embodiment, a square grid with a size of 5m×5m is preferred. Since the lowest point in a local area is usually the exposed ground surface, the corresponding point in three-dimensional space with the smallest Z-axis value is extracted in each block and recorded as the local ground reference point. Then, the vertical elevation difference and horizontal geometric distance between all other corresponding points in three-dimensional space in the block and the local ground reference point are calculated. The vertical elevation difference is divided by the horizontal geometric distance to obtain the local spatial gradient. In order to distinguish non-surface interference data from continuous natural steep slopes, this embodiment sets a natural terrain rest angle constant with a value of 1 based on the limit natural slope angle of 45° in conventional geological morphology. When the local spatial gradient is greater than the natural terrain rest angle constant, it is determined to be non-surface interference data and is removed.

[0097] After removing all non-surface interference data, data holes will remain in the original set of corresponding points in 3D space. To form a continuous terrain surface, these holes need to be filled. For each coordinate that generates a data hole, a search radius of 5m is set on the horizontal plane with that coordinate as the center. All local surface reference points falling within this search radius are extracted. The horizontal straight-line distance between the coordinate and each local surface reference point is calculated one by one. Based on the objective law that the natural surface is continuously and gradually changed in spatial distribution, that is, the closer the surface is, the more consistent its elevation is, the reciprocal of the calculated horizontal straight-line distance is used as the influence weight. After normalization, it is multiplied by the elevation of each local surface reference point, and then the results are summed. The resulting elevation is used to fill the position of the data hole, and finally a digital elevation model reflecting the true undulation characteristics of the exposed surface is obtained.

[0098] Subsequently, surface morphology indicators are extracted; the digital elevation model is cut into a series of square grids with equal side lengths on a horizontal two-dimensional plane. In this embodiment, 1m×1m is preferred. For each independent grid, its own elevation and the difference between the elevations of its eight adjacent grids are calculated. Each difference is divided by the geometric distance of the corresponding grid center point to obtain the local spatial gradient in eight adjacent directions. The arctangent angle of the local spatial gradient with the largest value is calculated. The angle value obtained is the absolute slope of the grid. The larger the absolute slope, the higher the risk of landslides and soil erosion caused by gravity shear stress.

[0099] Since curvature determines the convergence and divergence characteristics of surface runoff, the concave curvature of the water catchment area is prone to cause surface rainwater to converge and erode, exacerbating ecological erosion. Therefore, the system extracts the elevation change curve of adjacent grids along the direction of the largest elevation difference, and performs a second derivative on the curve. The second derivative obtained is the water catchment curvature.

[0100] Finally, ecological stability rules are established, and a dual joint determination is performed on each grid, as follows:

[0101] According to the ecological red line standard for steep slopes defined in the Soil and Water Conservation Law of the People's Republic of China, the critical slope is set at 25°. In spatial calculus logic, only when the curvature is greater than 0 will its geometric shape exhibit a concave funnel feature, causing rainwater from the surrounding surface to converge towards the center, forming a concentrated erosion runoff with destructive power.

[0102] Therefore, adjacent grids that simultaneously satisfy the conditions of absolute slope greater than critical slope and runoff curvature greater than 0 are merged to form multiple closed continuous regions. These regions, which are affected by both extremely steep terrain and runoff convergence and may suffer from severe soil erosion, are defined as ecological restoration zones. For each ecological restoration zone, its latitude and longitude coordinates are extracted and recorded to generate reconstruction instructions for subsequent construction guidance.

[0103] Figure 2 This is a comparison diagram of the effects of the prior art and the present invention provided in the embodiments of the present invention. It can be seen that when physical disturbance occurs, the prior art shows serious divergence deviation, while the present invention, due to the real-time overwriting and updating of the process noise covariance, always fits the reference elevation and has high accuracy and robustness.

Claims

1. A method for reconstructing national spatial ecological restoration zones based on remote sensing mapping, characterized in that, include: The system acquires and preprocesses real-time meteorological data and real-time electrical data of the target area to obtain a synchronized data sequence. The real-time meteorological data includes the real-time atmospheric pressure and real-time absolute temperature of the target area, and the real-time electrical data includes the real-time terminal voltage of the power battery of the mapping drone. A dynamic resonance boundary extrapolation model is constructed, and the synchronous data sequence is input into the model to calculate the dynamic resonance boundary frequency of the mapping UAV. in, For indexing time; For a moment The dynamic resonant boundary frequency; For a moment The aerodynamic resonant frequency; For electromechanical mapping factor; The nominal full-charge voltage of the power battery for the surveying drone; For a moment Real-time terminal voltage; The high-frequency acceleration data of the Z-axis of the mapping UAV is acquired, and the intensity of physical disturbance is extracted by frequency domain analysis of the high-frequency acceleration data of the Z-axis with the dynamic resonance boundary frequency as the upper limit of integration. Calculate the dynamic process noise covariance based on the physical disturbance intensity: in, For a moment The dynamic process noise covariance; The initial process noise covariance; This is the energy noise expansion factor; For a moment The intensity of physical disturbance; Update the dynamic process noise covariance to the preset process noise covariance matrix of the navigation system; The updated navigation system is used to extract exterior orientation elements, a digital elevation model is constructed based on the exterior orientation elements, and landform indicators are extracted using the digital elevation model. Then, ecological restoration zones are delineated, and reconstruction instructions are generated.

2. The method for reconstructing national spatial ecological restoration zones based on remote sensing mapping according to claim 1, characterized in that, The preprocessing includes: acquiring a second pulse hardware signal, using the rising edge of the second pulse hardware signal as an absolute zero timestamp, and using an interpolation algorithm to upsample the real-time meteorological data and the real-time electrical data, aligning them to a high-frequency timestamp consistent with the Z-axis high-frequency acceleration data, thereby obtaining the synchronization data sequence.

3. The method for reconstructing national spatial ecological restoration zones based on remote sensing mapping according to claim 1, characterized in that, The method further includes: obtaining the specific gas constant of the target area, and calculating the in-situ atmospheric density based on the real-time atmospheric pressure, the real-time absolute temperature, and the specific gas constant.

4. The method for reconstructing national spatial ecological restoration zones based on remote sensing mapping according to claim 1, characterized in that, The extraction of physical disturbance intensity includes: obtaining the acceleration power spectral density, and for each moment, taking 0 as the lower limit and the dynamic resonance boundary frequency at that moment as the upper limit, performing an integral operation on the acceleration power spectral density to obtain the physical disturbance intensity at that moment.

5. The method for reconstructing national spatial ecological restoration zones based on remote sensing mapping according to claim 1, characterized in that, Constructing a digital elevation model based on the exterior orientation elements specifically includes: acquiring remote sensing images of the target area; performing ray intersection operations on the remote sensing images carrying the exterior orientation elements to obtain corresponding points in three-dimensional space; removing non-surface interference data from the corresponding points in three-dimensional space; and fitting the data to generate the digital elevation model.

6. The method for reconstructing national spatial ecological restoration zones based on remote sensing mapping according to claim 1, characterized in that, The method further includes: performing gridding processing on the digital elevation model, extracting the absolute slope and drainage curvature of each grid as the landform index; dividing the target area into the ecological restoration zone according to the preset ecological stability rules and the landform index, and generating the reconstruction instruction with geographic coordinates.

7. A land spatial ecological restoration zoning reconstruction system based on remote sensing mapping, characterized in that, include: The processor and memory, wherein the memory stores computer program instructions, which, when executed by the processor, implement the method for reconstructing territorial spatial ecological restoration zones based on remote sensing mapping according to any one of claims 1-6.

Images