Deep and far sea unmanned aerial vehicle monitoring and management platform based on multi-sensor fusion

Abstract

The invention discloses a deep and far sea unmanned aerial vehicle monitoring and management platform based on multi-sensor fusion, and relates to the technical field of unmanned aerial vehicle marine monitoring. The intelligent analysis module is used for receiving the fusion data after time-space alignment and the real-time ocean current data, optimizing the moving path of the unmanned aerial vehicle according to the received data, evaluating the resource development possibility, obtaining an analysis result and transmitting the analysis result; an anti-interference communication module; an energy management module; and dynamically adjusting an energy distribution strategy. The module combines the task priority weight and the environment energy consumption weight, optimizes the energy use efficiency, prolongs the endurance time of the unmanned aerial vehicle, and ensures the smooth completion of the task. Particularly in a complex marine environment, navigation errors are corrected in real time through a quantum navigation error compensation model, the positioning precision is improved, and the moving path of the unmanned aerial vehicle is optimized in combination with real-time ocean current data and an ocean flow field cost function.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) ocean monitoring, and more specifically, to a deep - sea and far - sea UAV monitoring and management platform based on multi - sensor fusion. Background Art

[0002] At present, the application of UAVs in ocean monitoring still faces many technical challenges, such as difficult long - distance communication, insufficient endurance, and complex multi - sensor data fusion processing. To solve these problems, a deep - sea and far - sea UAV monitoring and management platform based on multi - sensor fusion has emerged. This platform integrates multiple sensors, advanced communication technologies, and intelligent management systems to achieve efficient and accurate monitoring and management of deep - sea and far - sea areas; UAVs can assist in ocean resource exploration, ocean engineering monitoring, etc., providing technical support for ocean resource development; Although UAVs have broad application prospects in ocean monitoring, the existing technologies still face many limitations: Communication technology limitations: In deep - sea and far - sea areas, the communication between UAVs and ground stations faces problems such as long distance, large signal attenuation, and many interference factors, resulting in poor communication stability and low data transmission rate; Insufficient endurance: Ocean monitoring tasks usually require long - term flight, but the existing UAVs have limited endurance and are difficult to meet the long - term monitoring needs of deep - sea and far - sea areas; Complex multi - sensor data fusion processing: Ocean monitoring involves multiple sensors, such as optical cameras, infrared cameras, sonars, etc. How to effectively fuse the data of these sensors to improve the monitoring accuracy and efficiency is a technical difficulty; Difficult navigation and positioning: In deep - sea and far - sea areas far from the coast, the GPS signal is weak or unstable, resulting in difficult navigation and positioning of UAVs; Insufficient environmental adaptability: The ocean environment is complex, with harsh conditions such as high salinity, high humidity, and strong winds and waves. There is still room for improvement in the environmental adaptability of existing UAVs. Summary of the Invention

[0003] To solve the above problems, the present invention provides a deep - sea and far - sea UAV monitoring and management platform based on multi - sensor fusion.

[0004] The deep - sea and far - sea UAV monitoring and management platform based on multi - sensor fusion includes A multi - source data acquisition module, which is used to carry sensors to collect multi - source data, and realizes the spatio - temporal alignment of multi - source data through GPS clock synchronization, and outputs the fused data after spatio - temporal alignment; An intelligent analysis module, which is used to receive the fused data after spatio - temporal alignment and real - time sea current data, then optimize the movement path of the UAV according to the received data, and evaluate the possibility of resource development, and obtain and transmit the analysis results; An anti-interference communication module, which is used to receive the collected multi-source data and analysis results, and then perform encrypted transmission and blockchain evidence storage on them; An energy management module, which is used to calculate quantum inertial navigation signals and dynamic energy scheduling strategies based on the collected multi-source data and analysis results; It is also used to provide positioning support for the mobile path planning results of the unmanned aerial vehicle according to the quantum inertial navigation signals.

[0005] Preferably, the multi-source data acquisition module includes a spatio-temporal alignment unit and a feature fusion unit: The spatio-temporal alignment unit is used to map the data from different sensors to the same time reference, providing a basis for subsequent data fusion; Specifically, the iterative closest point algorithm is used to match the point cloud data collected by the sonar sensor with the lidar data; The feature fusion unit adopts a lightweight model, calculates the attention weights of multiple sensors, and then performs weighted fusion on the multi-source data collected by multiple sensors according to the attention weights of multiple sensors to obtain the fused data after spatio-temporal alignment.

[0006] Preferably, the intelligent analysis module includes a dynamic path planning unit: The dynamic path planning unit adopts an improved RRT* algorithm, combines real-time sea current data and an ocean current field cost function to optimize the movement path of the unmanned aerial vehicle; During the path planning process, an ocean current field cost function C is introduced to evaluate the path cost; During each sampling and tree expansion process, real-time sea current data is obtained to update the parameters of the ocean current field cost function C to reflect the current ocean environmental conditions; According to the updated cost function C, the path costs of each node in the tree are re-evaluated, and the tree structure is adjusted to find a path with lower cost; Repeat the above steps until the target point is reached or the maximum number of iterations is satisfied to optimize the movement path of the unmanned aerial vehicle.

[0007] Preferably, the intelligent analysis module also includes a resource evaluation unit: The intelligent analysis module receives the fused data after spatio-temporal alignment and real-time sea current data; Obtain the historical resource database; Match the real-time fused data with the historical resource database, extract relevant mineral resource reserve information according to the model, and use the mineral resource reserve information as the analysis result to output the analysis result.

[0008] Preferably, the specific steps of the anti-interference communication module include: Receive the collected multi-source data and analysis results; Encrypt the received and collected multi-source data and analysis results; It also includes a blockchain evidence storage unit for storing the encrypted data on the blockchain. The specific steps are as follows: Perform data preparation and hashing, generating a unique hash value for the original data through a hashing algorithm; perform data packaging and signing, packaging the hash value and metadata into a transaction and signing the transaction with a private key; perform blockchain uploading, sending the transaction to the blockchain network, verifying and writing it into a block through a consensus mechanism to ensure that the evidence storage delay ≤ 2 seconds; perform evidence storage verification and publicity, verifying whether the hash value matches the original data and publicizing the evidence storage information through a blockchain browser.

[0009] Preferably, the specific working steps of the energy management module are as follows: Establish a quantitative relationship model between environmental parameters and energy consumption, defining environmental parameters, specifically including the salt spray corrosion factor, humidity influence coefficient, and wind and wave resistance energy consumption; Then calculate the comprehensive environmental adaptation coefficient based on the environmental parameters; Then adjust the energy allocation according to the environmental adaptation coefficient and task priority.

[0010] Preferably, the specific working steps of adjusting the energy allocation according to the environmental adaptation coefficient and task priority are as follows: Calculate the task priority weight WT of the drone by receiving the mineral resource reserve information; Then obtain the environmental energy consumption weight WK, which reflects the degree of restriction of the real-time environment on energy consumption; Finally, according to the formula ; Calculate the total energy allocation ratio of the drone .

[0011] Preferably, the steps of calculating the task priority weight WT of the drone by receiving the mineral resource reserve information are as follows: Obtain the mineral resource reserve parameter H in the mineral resource reserve information; it should be noted that through the above model: , calculate the mineral resource reserve parameter according to the model ; Then according to the formula , calculate and obtain the task priority weight .

[0012] Where is the theoretical maximum exploitable amount of this mineral resource reserve in the target area; obtained based on a geological model or industry standard; is the market price of this mineral, It is the highest market value among similar resources.

[0013] Preferably, the energy management module is further configured to correct navigation errors by fusing environmental data and improve the positioning accuracy under complex sea conditions. The specific steps are as follows: Receive the multi-source data collected, establish the error transfer function of the quantum gyroscope, and calculate its partial derivatives with respect to each environmental parameter; Input the change amount of environmental parameters, that is, the difference between the environmental parameters at the current moment and the previous moment and the calibrated partial derivatives. Multiply each change amount of environmental parameters by the corresponding partial derivative of the error transfer function, and then add all the results to obtain the error correction amount; Align the environmental parameter data and the navigation output according to the timestamp, and perform five-point moving average filtering on the change amount of environmental parameters to suppress high-frequency noise; Use FPGA hardware acceleration to calculate the error correction amount and the attitude correction amount. The processing delay is less than or equal to one millisecond. Inject the correction amount into the navigation solution loop, update the observation equation of the Kalman filter. The state equation describes the evolution of the system state, and the observation equation describes the corrected observation value, where the error correction amount is used as a correction term.

[0014] Beneficial effects: Through the multi-source data acquisition module, high-resolution data covering seabed topography, mineral resource distribution, and marine biological density are collected using devices such as optical sensors, sonar sensors, magnetometers, and CTD sensors. These data are time-space aligned through GPS clock synchronization and are processed in combination with the time-space alignment unit and the feature fusion unit. The time-space alignment unit uses the iterative closest point algorithm to eliminate the time-space reference differences between multi-source data. The feature fusion unit uses a lightweight model to calculate the attention weights of multi-sensors to achieve efficient data fusion. The fused data is used as the core input of the intelligent analysis module to trigger the resource reserve assessment and path planning functions. This multi-source data fusion technology significantly improves the accuracy and reliability of monitoring data, provides high-quality input for subsequent intelligent analysis and decision-making, and thus improves the monitoring accuracy and decision-making efficiency of the overall system; By establishing a quantitative relationship model between environmental parameters and energy consumption, the energy distribution strategy is dynamically adjusted. This module combines the task priority weight and the environmental energy consumption weight to optimize the energy use efficiency, extend the endurance time of the UAV, and ensure the successful completion of the task. Especially in complex marine environments, the navigation error is corrected in real time through the quantum navigation error compensation model to improve the positioning accuracy. The dynamic path planning unit uses an improved RRT* algorithm, combines real-time sea current data and the ocean current field cost function to optimize the movement path of the UAV. This dynamic adjustment mechanism not only improves the adaptability and task execution efficiency of the UAV in complex environments but also ensures the priority execution of high-priority tasks in case of energy shortage, thus enhancing the overall robustness and task success rate of the system. Brief Description of the Drawings

[0015] Figure 1 is the flowchart of the method of the present invention. Detailed Description of the Invention

[0016] As Figure 1 shown: The deep - sea and far - sea UAV monitoring and management platform based on multi - sensor fusion includes a multi - source data acquisition module, which is used to carry sensors to acquire multi - source data, and realizes the spatio - temporal alignment of multi - source data through GPS clock synchronization, and outputs the fused data after spatio - temporal alignment; It should be noted that, in this embodiment, the carried sensors include the original data of an optical sensor (multi - spectral camera, wavelength 400 - 1000nm), a sonar sensor (multi - beam bathymeter, detection depth ≥ 1000m), a magnetometer (sensitivity ≤ 1nT), and a thermosalinograph (salinity error ≤ 0.002PSU); Optical sensor Acquisition content: The multi - spectral camera acquires optical image data of different bands in the wavelength range of 400 - 1000nm, which can reflect information such as the radiation characteristics, surface texture, and color of objects; Acquisition method: Through the optical lens for focusing imaging, the photosensitive element converts the optical image into an electrical signal, and then through steps such as analog - to - digital conversion and digital processing, digital image data that can be stored and transmitted is obtained; Sonar sensor Acquisition content: The multi - beam bathymeter is used to measure the water depth at the bottom, and can also obtain information such as the underwater topographic features, geological structure, and obstacles such as sunken ships and reefs; Acquisition method: Transmit acoustic pulses, the acoustic waves propagate in water and are reflected when encountering the interface of different media, receive the reflected wave signals, and calculate the water depth and underwater topography and other information according to the acoustic wave propagation time and speed; Magnetometer: Acquisition content: The magnetometer with a sensitivity ≤ 1nT can measure the minute changes in the earth's magnetic field, which can reflect information such as the magnetic differences of underground rocks, the distribution of mineral resources, and archaeological remains, and is widely used in fields such as geological exploration, mineral exploration, and archaeological detection; Acquisition method: Through principles such as induction coils or magnetoresistive effects, convert the magnetic field changes into electrical signals, and through processing such as amplification, filtering, and analog - to - digital conversion, obtain the magnetic field intensity data; Thermosalinograph: Acquisition content: Measure the temperature, salinity, and depth of seawater, with a salinity error ≤ 0.002PSU; Collection method: Temperature is measured by sensors such as thermistors; salinity is calculated based on the conductivity measurement principle, combined with temperature and pressure data; depth is obtained by converting the pressure measured by the pressure sensor according to the relationship with water depth.

[0017] The intelligent analysis module is used to receive the fused data after spatio-temporal alignment and real-time ocean current data, then optimize the movement path of the UAV according to the received data, evaluate the possibility of resource development, obtain the analysis results and transmit them; It should be noted that the spatio-temporal alignment fused data generated by the multi-source data collection module (resolution ≤ 1m, covering seabed topography, distribution of mineral resources, and density of marine organisms) is one of the core inputs of the intelligent analysis and decision-making module; The fused data triggers the resource reserve assessment and path planning functions of the intelligent analysis module; The intelligent analysis module combines the historical resource database and real-time ocean current data (measured by ADCP, accuracy ±0.05m / s), evaluates the feasibility of resource development through reinforcement learning path optimization planning and Bayesian network, and generates a mineral resource reserve assessment report and dynamic path planning instructions; The anti-interference communication module is used to receive the multi-source data and analysis results collected, and then perform encrypted transmission and blockchain certification on them; The energy management module is used to calculate the quantum inertial navigation signal and dynamic energy scheduling strategy according to the multi-source data and analysis results collected; It is also used to provide positioning support for the movement path planning result of the UAV according to the quantum inertial navigation signal.

[0018] It should be noted that the collection module generates fused data to the analysis module to trigger resource assessment and path planning to the communication module to encrypt and transmit the results to the intelligent analysis module to evaluate the resource development model.

[0019] Preferably, the multi-source data collection module includes a spatio-temporal alignment unit and a feature fusion unit: The spatio-temporal alignment unit is used to map the data from different sensors to the same time reference, providing a basis for subsequent data fusion; Specifically, the iterative closest point algorithm is used to match the point cloud data collected by the sonar sensor with the lidar data; The ICP algorithm iteratively calculates the closest point pairs between two groups of point clouds and adjusts their relative positions and postures until the set residual threshold (≤0.1m) is reached; this matching method can effectively eliminate the spatio-temporal reference differences between multi-source data and improve the accuracy and efficiency of data fusion; For other data, the timestamp alignment method can be used to unify the timestamps of each sensor data to the GPS time reference, and then adjust the data time through interpolation or extrapolation algorithms to achieve spatio-temporal alignment; The feature fusion unit adopts a lightweight model to calculate the attention weights of multiple sensors, and then performs weighted fusion on the multi-source data collected by the multiple sensors according to the attention weights of the multiple sensors to obtain the fused data after spatio-temporal alignment.

[0020] It should be noted that in this embodiment, the steps for calculating the attention weights of multiple sensors are as follows: 1. Among them 、 and T respectively represent the query, key, and value matrices. In this embodiment, , in this way, the model can effectively capture the correlation between different sensor data, and the calculated is the weight of the th sensor, where V1 is a matrix obtained by linearly transforming the original sensor data, and its essence is the mapping result of the original sensor information in the feature space; Preferably, the intelligent analysis module includes a dynamic path planning unit: The dynamic path planning unit adopts an improved RRT* algorithm, combines real-time sea current data and an ocean current field cost function, and optimizes the movement path of the UAV; The improved RRT* algorithm includes: initializing the starting point and the target point , taking the starting point as the root node and establishing a tree T that only contains the starting point; Setting the neighborhood radius r for finding candidate nodes around the root node for optimization; Randomly generating a point from the map as the sampling point, which is in the free space and not in the obstacle; Finding the nearest neighbor node in the tree T, and calculating whether the path from the nearest neighbor node to the point collides with the obstacle; if there is no collision, generating a new node and connecting it to ; Finding all the nodes within the neighborhood, that is, the nodes whose paths are less than the radius r; Finding a shortest and collision-free path from these nodes to connect , if the cost of this path is lower, updating the parent node of to the optimal node within this neighborhood and adding the new edge to the tree; After is added to the tree, checking other nodes within its neighborhood to determine whether to pass through Connecting these nodes can shorten their path lengths. If possible, update the parent nodes of these nodes to and adjust the corresponding paths; During each sampling and tree expansion process, introduce the ocean current field cost function C to evaluate the path cost; It should be noted that the specific steps of the ocean current field cost function C are as follows: ; where represents the velocity vector of the ocean current; represents the time step, which is determined by the iteration frequency of the path planning algorithm. In this embodiment, it is a fixed parameter of 1 second. represents the risk assessment value of the path, which is calculated by the following method: Obstacle distance: Calculate the minimum distance between the path and the obstacle. The closer the distance, the higher the risk; Flow field intensity: Evaluate the risk according to the magnitude and direction of the flow field velocity. The greater the flow field intensity, the higher the risk; Path complexity: Consider the tortuosity of the path and the distribution of obstacles. The higher the complexity, the higher the risk; Obstacle distance: Calculate the minimum distance between the path and the obstacle. The closer the distance, the higher the risk of the UAV colliding. Therefore, it will occupy a larger weight in the risk assessment value. Usually, a distance sensor or a geometric calculation method based on map information can be used to determine the distance between the path and the obstacle.

[0021] Flow field intensity: Evaluate the risk according to the magnitude and direction of the flow field velocity. The greater the flow field intensity, the greater the impact on the flight stability of the UAV, which may cause the UAV to deviate from the predetermined route or increase the difficulty of flight control, thereby increasing the risk assessment value. The flow field intensity can be determined by the flow velocity information in the real-time obtained ocean current data and comprehensively evaluated in combination with the relationship between the direction of the flow field and the flight direction of the UAV.

[0022] Path complexity: Consider the tortuosity of the path and the distribution of obstacles. The more tortuous the path, it means that the UAV needs to make more turns and speed adjustments, which will increase the complexity of flight control and the probability of errors. At the same time, the distribution of obstacles will also affect the safety of the path. The risk in the area with dense obstacles is relatively high. The path complexity can be evaluated through the geometric shape analysis of the path and the obstacle detection algorithm and incorporated into the calculation of the risk assessment value; When determining the risk assessment value, corresponding weights need to be assigned to these three factors according to the specific path planning problem and application scenario. In this embodiment, the initial weights of the three factors are one-third; During each sampling and expansion of the tree, ocean current data is obtained in real time to update the parameters of the ocean current field cost function C, so as to reflect the current ocean environmental conditions; According to the updated cost function C, re-evaluate the path costs of each node in the tree and adjust the tree structure to find a path with lower cost; Repeat the above steps until the target point is reached or the maximum number of iterations is satisfied, and optimize the movement path of the UAV.

[0023] It should be noted that the initial path is optimized to reduce the turning points in the path, making the path smoother and more direct. Starting from the starting point, check whether there are obstacles in the connection lines between each node in the path in turn. If there are no obstacles, simplify the path; According to the real-time ocean current data and environmental changes, dynamically adjust the path planning strategy to ensure the optimality and feasibility of the path.

[0024] Output the optimized path planning result in JSON format, including information such as the movement path of the UAV and task priorities.

[0025] Preferably, the intelligent analysis module further includes a resource evaluation unit: The intelligent analysis module receives the fusion data after spatio-temporal alignment and the real-time ocean current data; The multi-spectral camera analyzes the mineral composition of the seabed surface layer through multi-spectral bands. For example, iron oxides have characteristic absorption peaks in the near-infrared band (700 - 900nm), which can be used to preliminarily delineate the mineralized area. Combining with the spectral feature library in the historical database, match the mineral types and generate a grade distribution map, and use radiation correction and atmospheric correction algorithms to eliminate the influence of underwater light attenuation and generate a reflectivity map.

[0026] The multi-beam echosounder constructs a three-dimensional seabed terrain model, calculates the volume of the ore body, and identifies the shape and distribution boundary of the ore body through the coverage width (more than 7 times the water depth) and high precision (IHO special class standard) scanning of the multi-beam sonar; Use interpolation algorithms (such as Kriging method) to fill in the data missing areas, and calculate the volume of the ore body in combination with the digital elevation model (DEM); The magnetometer detects the seabed magnetic anomaly and infers the distribution density of iron-containing minerals (such as magnetite). The magnetic gradient data can assist in identifying hidden ore bodies; Apply dynamic calibration algorithms (such as real-time offset correction during UAV flight) to eliminate geomagnetic interference and improve the signal-to-noise ratio of the data; The thermosalinograph monitors the influence of seawater salinity and temperature on the chemical stability of the ore body. For example, high salinity may accelerate the oxidation of sulfide deposits, and an environmental correction factor needs to be introduced in the reserve calculation; The critical parameters include the ore body density, obtained by matching the magnetic anomaly intensity with the historical core sample database, the volume, obtained by segmenting the ore body blocks based on the multibeam terrain model, and the grade function, obtained by establishing a non-linear relationship using the multispectral reflectance and the mineral spectral feature library; Obtain the historical resource database; it should be noted that in this embodiment, the historical resource database is obtained from the China National Geological Archives of Data; Match the real-time fusion data with the historical resource database, extract relevant mineral resource reserve information according to the model, and take the mineral resource reserve information as the analysis result and output the analysis result.

[0027] In this embodiment, the model is specifically: , calculate the mineral resource reserve parameters according to the model , and take it as the historical mineral resource reserve information; where is the ore body density obtained by matching the magnetic anomaly intensity with the historical core sample database; is the ore body volume, obtained by segmenting the ore body blocks based on the multibeam terrain model, is the grade function, obtained by establishing a non-linear relationship using the multispectral reflectance and the mineral spectral feature library.

[0028] Preferably, the specific steps of the anti-interference communication module include: Receive the collected multi-source data and the analysis result; Perform encryption processing on the received collected multi-source data and the analysis result; the specific steps are as follows: Select an encryption algorithm, select a symmetric encryption, asymmetric encryption or hybrid encryption algorithm according to the data type and security requirements; perform key management, including key generation, storage and distribution, to ensure the security of the key; perform data encryption, encrypt static data and dynamic data, and implement transparent encryption; perform integrity verification and access control, use a hash algorithm or a message authentication code to ensure data integrity, and implement multi-factor authentication and the least privilege access policy; perform auditing and optimization regularly, update the encryption algorithm, and monitor the encryption performance; It also includes a blockchain evidence storage unit for performing blockchain evidence storage on the encrypted data, and the specific steps are as follows: Perform data preparation and hashing processing to generate a unique hash value for the original data through a hashing algorithm; perform data packaging and signing, package the hash value and metadata into a transaction, and sign the transaction using the private key; perform blockchain uploading, send the transaction to the blockchain network, verify and write it into a block through a consensus mechanism to ensure that the deposit delay ≤ 2 seconds; perform deposit verification and publicity, verify whether the hash value matches the original data, and publicize the deposit information through a blockchain browser.

[0029] Preferably, the specific working steps of the energy management module are as follows: Establish a quantitative relationship model between environmental parameters and energy consumption, define environmental parameters, specifically including the salt spray corrosion factor, humidity influence coefficient, and wind and wave resistance energy consumption; It should be noted that in this embodiment, the salt spray corrosion factor CS is obtained by dividing the collected real-time salinity by the standard salinity, and the standard salinity is 35 PSU; The humidity influence coefficient is H2, which is obtained through where is the collected real-time humidity, which takes effect when the humidity is greater than 80%, otherwise it takes a value of 0; The wind and wave resistance energy consumption PW is obtained by multiplying 0.5 by the air density, the square of the wind speed, the cross-sectional area of the drone facing the wind, and the wind resistance coefficient; Then calculate the comprehensive environmental adaptation coefficient according to the environmental parameters; In this embodiment, the specific steps are: ; where is the reference power consumption; Then adjust the energy allocation according to the environmental adaptation coefficient and the task priority.

[0030] Preferably, the specific working steps of adjusting the energy allocation according to the environmental adaptation coefficient and the task priority are as follows: Calculate the task priority weight WT of the drone by receiving the mineral resource reserve information; Then obtain the environmental energy consumption weight WK, which reflects the degree of restriction of the real-time environment on energy consumption; It should be noted that the environmental energy consumption weight WK is a parameter used to measure the impact of environmental conditions on energy consumption, and its calculation formula is 1 minus KV: where KV is the environmental energy consumption coefficient, which reflects the degree of restriction of the environment on energy consumption.

[0031] KV represents the degree of influence of environmental conditions on energy consumption. This coefficient can be obtained from the linear influence of real-time environmental monitoring data (such as temperature, humidity, pollutant concentration, etc.) on energy consumption; By introducing the environmental energy consumption weight, the energy allocation strategy can be dynamically adjusted to ensure the reasonable allocation of energy under different environmental conditions, avoiding energy waste or shortage caused by environmental factors; Finally, according to the formula ; Calculate the total energy allocation ratio of the drone .

[0032] It should be noted that to determine the energy allocation ratio between different tasks or regions, the energy allocation ratio combines the task priority and the environmental energy consumption weight to achieve the reasonable allocation of energy. When the environmental energy consumption is high, the total allocation ratio will be correspondingly reduced, so as to give priority to the execution of high-priority tasks in the case of energy shortage; In the UAV path planning, the energy allocation can be dynamically adjusted according to the environmental conditions and task priorities to ensure the efficient operation of the UAV in complex environments; In the energy management module, through real-time monitoring and adjustment, the energy use efficiency is optimized, the endurance time of the UAV is extended, and the successful completion of the task is ensured.

[0033] Preferably, the steps of calculating the task priority weight WT of the drone by receiving the mineral resource reserve information are as follows: Obtain the mineral resource reserve parameter H in the mineral resource reserve information; it should be noted that through the above model: , calculate the mineral resource reserve parameter according to the model ; Then according to the formula , calculate and obtain the task priority weight .

[0034] Where is the theoretical maximum exploitable amount of this mineral resource reserve in the target area; obtained based on the geological model or industry standards; is the market price of this mineral, is the highest market value of similar resources.

[0035] It should also be noted that in marine resource exploration, the UAV can dynamically adjust its path and task execution order according to the task priority weight. For example, when detecting a mineral area with high reserves and high value, the UAV will preferentially allocate more resources and time for detailed exploration, while low-priority areas can be quickly skimmed over or processed later. This strategy not only improves the efficiency of resource exploration but also ensures the priority completion of key tasks.

[0036] Preferably, the energy management module is further configured to correct the navigation error by fusing environmental data to improve the positioning accuracy under complex sea conditions. The specific steps are as follows: Receive the multi-source data collected, establish the error transfer function of the quantum gyroscope, and calculate its partial derivatives with respect to each environmental parameter; the partial derivative represents the rate of change of the error transfer function with respect to a certain environmental parameter and is obtained through limit calculation; Among them, the multi-source data is the multi-source data collected by the multi-source data acquisition module. Here, the "its" in "its partial derivatives with respect to each environmental parameter" refers to the error transfer function of the quantum gyroscope, and each partial derivative represents the sensitivity of the error transfer function to a certain environmental parameter; The error transfer function is , where is the angular velocity error, is the environmental parameter, which is temperature, magnetic field, vibration, and air pressure in sequence; In actual implementation, measure the error change through offline calibration (control variable method) and fit to obtain the empirical formula of the partial derivative; Input the environmental parameter change amount, that is, the difference between the environmental parameters at the current moment and the previous moment and the calibrated partial derivative, multiply each environmental parameter change amount by the corresponding partial derivative of the error transfer function, and then add all the results to obtain the error correction amount; Output the position correction amount and the attitude correction amount (the error correction amount divided by the characteristic length of the UAV); Align the environmental parameter data and the navigation output according to the time stamp, perform five-point moving average filtering on the environmental parameter change amount to suppress high-frequency noise; Use FPGA hardware acceleration to calculate the error correction amount and the attitude correction amount, the processing delay is less than or equal to one millisecond, inject the correction amount into the navigation solution loop, and update the observation equation of the Kalman filter. The state equation describes the evolution of the system state, and the observation equation describes the corrected observation value, where the error correction amount is used as the correction term. It can suppress the divergence of the state estimation caused by environmental mutations and reduce the positioning error variance.

[0037] It should be noted that the main task of the error correction calculation unit is to receive data from different sensors, such as the angular velocity of the IMU, the values of the accelerometer, and the environmental parameter change amount, etc. Subsequently, with the help of the empirical formula of the partial derivative obtained through offline calibration, calculate the error correction amount. Specifically, the calculation of the error correction amount is based on the environmental parameter change amount (such as ΔT, ΔB) and the calibrated partial derivative (such as ∂E / ∂T, ∂E / ∂B), and is completed by linear superposition. Taking the magnetic field correction amount as an example, its calculation method is to multiply ΔB by the magnetic field sensitivity ∂E / ∂B. To ensure the applicability and accuracy of the calculation result, it is also necessary to normalize the error correction amount, that is, divide it by the characteristic length L of the UAV to obtain the attitude correction amount Δθ; The attitude calculation unit is based on the quaternion or equivalent rotation vector model to fuse the sensor data corrected for errors, thereby accurately calculating the attitude angles, including the pitch angle, roll angle, and yaw angle. In this process, complementary filtering or dynamic gradient descent method may be used to suppress dynamic interference and improve the accuracy of the attitude angle correction amount. For example, the gyroscope data corrected for errors is used to drive the quaternion differential equation and combined with the observations of the accelerometer to achieve attitude calculation. In addition, the error correction amount can be used as the observation deviation to update the posterior estimate of the equivalent rotation vector, further optimizing the attitude calculation result. The Kalman filter update unit is responsible for dynamically adjusting the state equation and observation equation and effectively injecting the correction amount into the observation link of the filter. In this process, the state equation is used to describe the dynamic evolution process of the system state, usually presented as a non-linear model, and its general form is x_{k + 1}=f(x_k,u_k)+w_k, where x_k represents the system state, such as attitude angle, angular velocity, position, etc.; u_k is the control input, such as the angular velocity of the gyroscope; w_k represents the process noise. The observation equation is used to establish the relationship between the corrected sensor observations and the system state, and its form is z_k=h(x_k)+v_k, where z_k is the IMU data including error correction, such as the measurement value of the accelerometer; v_k is the observation noise. There are mainly two ways to inject the correction amount: one is the observation value correction, that is, the error correction amount is used as the compensation term of the observation equation. For example, the attitude correction amount Δθ is added to the measurement value z_k of the accelerometer to obtain the corrected observation value z_k'=z_k+Δθ, and then the Kalman gain is calculated using the corrected observation value, thereby optimizing the state estimate; the other is the direct correction of the state vector. After the prediction stage, the error correction amount is directly added to the state estimate value, and the formula is x_{k|k}=x_{k|k - 1}+K_k(z_k - h(x_{k|k - 1}))+Δθ, where K_k is the Kalman gain. The FPGA realizes complex operations such as covariance matrix update and Jacobian matrix linearization through custom logic, thus effectively reducing the calculation latency. For example, in the Extended Kalman Filter (EKF), the formula for covariance matrix update is: Pk∣k=(I−KkH)Pk∣k−1 where Pk∣k is the updated covariance matrix, I is the identity matrix, Kk is the Kalman gain, H is the observation matrix, and Pk∣k−1 is the covariance matrix in the prediction stage. Jacobian matrix linearization involves performing a Taylor expansion on the nonlinear system model and taking the first-order approximation to obtain the linearized state transition matrix and observation matrix. At the same time, the pipelining and time-division multiplexing technologies are adopted to split the prediction-update process of the Kalman filter into multiple pipeline stages, and each stage is responsible for specific calculation tasks, such as system state prediction and covariance matrix prediction in the prediction stage, Kalman gain calculation, state update, and covariance matrix update in the update stage, etc. In this way, the parallel processing ability of the FPGA can be fully utilized to improve the calculation efficiency. In addition, the fixed-point number optimization technology is also used to quantize the model parameters into fixed-point numbers. For example, the elements of the covariance matrix are quantized into the fixed-point number format, which reduces the resource occupancy and improves the operation speed while ensuring that the calculation accuracy meets the requirements; The above is only the preferred implementation mode of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions falling within the idea of the present invention belong to the protection scope of the present invention. It should be pointed out that for those of ordinary skill in the art, several improvements and refinements made without departing from the principle of the present invention should also be regarded as within the protection scope of this template.

Claims

1. A deep - sea and far - sea unmanned aerial vehicle monitoring and management platform based on multi - sensor fusion, characterized in that, including a multi-source data acquisition module, which is used to carry sensors to collect multi-source data, and realizes the spatio-temporal alignment of multi-source data through GPS clock synchronization, and outputs the fused data after spatio-temporal alignment; an intelligent analysis module, which is used to receive the fused data after spatio-temporal alignment and real-time sea current data, then optimize the moving path of the UAV according to the received data, and evaluate the possibility of resource development, and obtain and transmit the analysis results; an anti-interference communication module, which is used to receive the collected multi-source data and analysis results, and then perform encrypted transmission and blockchain evidence storage on them; an energy management module, which is used to calculate the quantum inertial navigation signal and dynamic energy scheduling strategy according to the collected multi-source data and analysis results; It is also used to provide positioning support for the moving path planning result of the UAV according to the quantum inertial navigation signal.

2. The deep - sea and far - sea UAV monitoring and management platform based on multi - sensor fusion according to claim 1, characterized in that, The multi-source data acquisition module includes a spatio-temporal alignment unit and a feature fusion unit: The spatio-temporal alignment unit is used to map the data from different sensors to the same time reference, providing a basis for subsequent data fusion; Specifically, the iterative closest point algorithm is used to match the point cloud data collected by the sonar sensor with the lidar data; The feature fusion unit adopts a lightweight model to calculate the attention weights of multiple sensors, and then performs weighted fusion on the multi-source data collected by multiple sensors according to the attention weights of multiple sensors to obtain the fused data after spatio-temporal alignment.

3. The deep-sea and far-sea unmanned aerial vehicle monitoring and management platform based on multi-sensor fusion according to claim 1, characterized in that The intelligent analysis module includes a dynamic path planning unit: The dynamic path planning unit adopts an improved RRT* algorithm, combines real-time sea current data and ocean current field cost function to optimize the moving path of the UAV; During the path planning process, the ocean current field cost function C is introduced to evaluate the path cost; During each sampling and expanding the tree process, real-time sea current data is obtained to update the parameters of the ocean current field cost function C to reflect the current ocean environmental conditions; According to the updated cost function C, re-evaluate the path costs of each node in the tree, adjust the tree structure to find a path with lower cost; Repeat the above steps until the target point is reached or the maximum number of iterations is satisfied, and optimize the moving path of the UAV.

4. The deep-sea and far-sea UAV monitoring and management platform based on multi-sensor fusion according to claim 3, characterized in that, The intelligent analysis module also includes a resource evaluation unit: The intelligent analysis module receives the fused data after spatio-temporal alignment and real-time sea current data; Obtain the historical resource database; Match the real-time fused data with the historical resource database, extract relevant mineral resource reserve information according to the model, and use the mineral resource reserve information as the analysis result to output the analysis result.

5. The deep - sea and far - sea UAV monitoring and management platform based on multi - sensor fusion according to claim 4, characterized in that, The specific steps of the anti-interference communication module include: Receive the collected multi-source data and analysis results; Perform encryption processing on the received collected multi-source data and analysis results; It also includes a blockchain evidence storage unit, which is used to perform blockchain evidence storage on the encrypted data. The specific steps are as follows: Perform data preparation and hashing, generate a unique hash value for the original data through a hashing algorithm; perform data packaging and signing, package the hash value and metadata into a transaction, and sign the transaction using a private key; perform blockchain uploading, send the transaction to the blockchain network, verify it through a consensus mechanism and write it into a block to ensure that the deposit delay ≤ 2 seconds; perform deposit verification and publicity, verify whether the hash value matches the original data, and publicize the deposit information through a blockchain browser.

6. The deep - sea and far - sea UAV monitoring and management platform based on multi - sensor fusion according to claim 4, characterized in that, The specific working steps of the energy management module are as follows: Establish a quantitative relationship model between environmental parameters and energy consumption, define environmental parameters, specifically including the salt spray corrosion factor, humidity influence coefficient, and wind and wave resistance energy consumption; Then calculate the comprehensive environmental adaptation coefficient according to the environmental parameters; Then adjust the energy allocation according to the environmental adaptation coefficient and task priority.

7. The deep - sea and far - sea UAV monitoring and management platform based on multi - sensor fusion according to claim 6, characterized in that, The specific working steps of adjusting the energy allocation according to the environmental adaptation coefficient and task priority are as follows: Calculate the task priority weight WT of the UAV by receiving the mineral resource reserve information; Then obtain the environmental energy consumption weight WK, which reflects the degree of restriction of the real-time environment on energy consumption; Finally, according to the formula ; Calculate the total energy distribution ratio of the UAV .

8. The deep - sea and far - sea UAV monitoring and management platform based on multi - sensor fusion according to claim 7, characterized in that, The steps of calculating the task priority weight WT of the UAV by receiving the mineral resource reserve information are as follows: Obtain the mineral resource reserve parameter H in the mineral resource reserve information; it should be noted that through the above model: , mineral resource reserve parameters are calculated according to the model ; Then, according to the formula , calculate to obtain the task priority weight ; Among them is the theoretical maximum extractable amount of the mineral resource reserves in the target area; obtained based on the geological model or industry standards; is the market price of this mineral resource, is the highest market value of similar resources.

9. The deep - sea and far - sea UAV monitoring and management platform based on multi - sensor fusion according to claim 8, characterized in that, The energy management module is also used to fuse environmental data to correct navigation errors and improve the positioning accuracy under complex sea conditions. The specific steps are as follows: Receive the collected multi-source data, establish a quantum gyroscope error transfer function, and calculate its partial derivatives with respect to each environmental parameter; Input the environmental parameter change amount, that is, the difference between the environmental parameters at the current moment and the previous moment and the calibrated partial derivatives, multiply each environmental parameter change amount by the corresponding error transfer function partial derivative, and then add all the results to obtain the error correction amount; Align the environmental parameter data and the navigation output according to the timestamp, perform five-point moving average filtering on the environmental parameter change amount to suppress high-frequency noise; Use FPGA hardware acceleration to calculate the error correction amount and attitude correction amount, with a processing delay less than or equal to one millisecond. Inject the correction amount into the navigation solution loop to update the observation equation of the Kalman filter. The state equation describes the evolution of the system state, and the observation equation describes the corrected observation value, where the error correction amount is used as a correction term.

Images