Unmanned aerial vehicle flight risk assessment and quality monitoring method

Abstract

The invention relates to the technical field of unmanned aerial vehicle electronic data processing, and provides an unmanned aerial vehicle flight risk assessment and quality monitoring method, which comprises the following steps of: acquiring diversified data acquired in a flight process of an unmanned aerial vehicle, establishing a data optimization group, performing optimization correction, extracting features and constructing a risk assessment model; real-time flight monitoring data is classified and corrected and then substituted into the model, and a comprehensive flight risk assessment result is obtained. The problems of sensor data errors and noise are solved, the evaluation accuracy is improved, decision support is provided for safe flight of the unmanned aerial vehicle, the flight safety and efficiency of the unmanned aerial vehicle are remarkably enhanced, and the flight risk evaluation and quality monitoring accuracy of the unmanned aerial vehicle is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle electronic data processing, and in particular to a method for unmanned aerial vehicle flight risk assessment and quality monitoring. Background Art

[0002] Drone flight risk assessment is a comprehensive process involving multiple aspects, including the performance, operating environment, potential risks, and regulatory compliance of drones. Drone performance assessment includes comprehensive inspection and testing of drone hardware and flight performance, thereby assessing the various potential risks faced by drones during flight, including crash risk, collision risk, data leakage risk, etc. Hardware equipment testing includes checking the integrity and functionality of key components such as motors, propellers, batteries, and cameras, so that drones will not cause accidents due to hardware failures during flight. Flight performance testing includes testing the maximum load, maximum flight altitude, maximum flight speed, and endurance of drones to understand the performance limits and applicable scenarios of drones.

[0003] UAV flight quality monitoring is a key link to ensure that UAVs maintain efficient, stable and safe operation when performing tasks. UAVs need to be monitored in real time during flight to ensure that they fly according to the predetermined trajectory and parameters. UAV flight quality monitoring includes real-time tracking of the UAV's position through GPS or other positioning systems to ensure that the UAV does not deviate from the predetermined route. Monitoring the UAV's flight speed, altitude, attitude, battery power and other status parameters to promptly detect and handle abnormal conditions of the UAV.

[0004] Drone flight risk assessment and quality monitoring are two complementary links. Together, they constitute an important guarantee for the safe flight of drones. Combining drone flight risk assessment with quality monitoring can more effectively prevent flight accidents and improve flight efficiency. Drone flight risk assessment is the process of identifying, analyzing and evaluating potential risks before and during flight. Through risk assessment, various risk factors that drones may encounter during flight can be identified, such as bad weather, airspace restrictions, equipment failures, etc.

[0005] The flight risk assessment and quality monitoring of drones during flight require real-time collection of various sensor data, including position, speed, attitude, environmental parameters, etc. However, there are errors in the data collected by the sensors during the flight of drones in the air, and there is also the introduction of noise data during the transmission of sensor data, which affects the availability of sensor data and reduces the accuracy of drone flight risk assessment and quality monitoring. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention provides a method for unmanned aerial vehicle flight risk assessment and quality monitoring, which solves the problem that errors exist in the data collected by sensors during the flight of unmanned aerial vehicles, and noise data is introduced during the transmission of sensor collected data, thereby affecting the availability of sensor collected data and reducing the accuracy of unmanned aerial vehicle flight risk assessment and quality monitoring.

[0007] In order to solve the above technical problems, the specific technical solutions of the present invention are as follows: The present invention provides a method for unmanned aerial vehicle flight risk assessment and quality monitoring, comprising: Step S101, acquiring data collected during the flight of the drone, the data collected during the flight of the drone including image and video data, infrared thermal imaging data, spectral data, laser scanning data, inertial measurement data, barometer data, position data, meteorological data, radiation measurement data and sonar data, the image and video data including night image and video data and day image and video data; Step S102, establishing data optimization groups, the data optimization groups include a first data optimization group, a second data optimization group, a third data optimization group, a fourth data optimization group, a fifth data optimization group, a sixth data optimization group and a seventh data optimization group, copying the position data to generate first position data, second position data and third position data, copying the spectral data to generate first spectral data and second spectral data, classifying night image video data and infrared thermal imaging data into the first data optimization group, classifying day image video data and the first spectral data into the second data optimization group, classifying laser scanning data and the first position data into the third data optimization group, classifying inertial measurement data and barometer data into the fourth data optimization group, classifying the second position data and meteorological data into the fifth data optimization group, classifying radiation measurement data and the second spectral data into the sixth data optimization group, and classifying sonar data and the third position data into the seventh data optimization group; Step S103, optimizing and correcting the data in the first data optimization group, the second data optimization group, the third data optimization group, the fourth data optimization group, the fifth data optimization group, the sixth data optimization group, and the seventh data optimization group in sequence, to obtain corrected data collected during the flight of the drone; Step S104, extracting data features from the data correction results of the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping, and the seventh data optimization grouping to obtain height error features, flight attitude stability features, meteorological condition suitability features, and obstacle distance features, performing feature fusion on the height error features, flight attitude stability features, meteorological condition suitability features, and obstacle distance features to obtain risk assessment indicator feature fusion data, using the data collected during historical UAV flights and historical risk assessment results as training sets, training and verifying the random forest model, using the height error, flight attitude stability, meteorological condition suitability, obstacle distance, etc. as input features of the random forest model, and obtaining a UAV flight risk assessment model; Step S105, obtaining the real-time flight monitoring data of the UAV, classifying the real-time flight monitoring data of the UAV according to the data characteristics, obtaining the classification results of the characteristics of the real-time flight monitoring data of the UAV, grouping the classification results of the characteristics of the real-time flight monitoring data of the UAV, and correcting the characteristics of the grouped real-time flight monitoring data of the UAV to obtain the corrected real-time flight monitoring data characteristics of the UAV, substituting the corrected real-time flight monitoring data characteristics of the UAV into the UAV flight risk assessment model to obtain the comprehensive flight risk assessment results of the UAV, which include the risk assessment results of the altitude error, the risk assessment results of the flight attitude stability, the risk assessment results of the weather conditions suitability and the risk assessment results of the obstacle distance.

[0008] Furthermore, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S101: Establish data preprocessing batches according to the type and source of data collected during the drone flight, store the data collected during the drone flight in the blockchain in sequence according to the data preprocessing batches, and generate hash values ​​corresponding to the data collected during the drone flight; Preprocess the data collected during the flight of the drone, store the data generated during the preprocessing of the data collected during the flight of the drone and the data collected during the flight of the drone after the preprocessing is completed in the blockchain, and generate the corresponding hash value; When receiving the instruction to call the data information collected during the flight of the drone, recalculate the hash value corresponding to the data collected during the flight of the called drone, and match the calculation result with the original hash value of the data collected during the flight of the drone; If the hash values ​​match, the data collected during the flight of the called drone has not been tampered with. If the hash values ​​do not match, the data collected during the flight of the called drone has been tampered with.

[0009] Furthermore, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S102: receiving a target of data optimization grouping, and determining a data type to be classified according to the target of data optimization grouping, wherein the data type to be classified includes night image video data, day image video data, infrared thermal imaging data, first spectrum data, second spectrum data, laser scanning data, first position data, second position data, third position data, inertial measurement data, barometer data, meteorological data, radiation measurement data, and sonar data; Perform correlation analysis on night image video data, day image video data, infrared thermal imaging data, first spectrum data, second spectrum data, laser scanning data, first position data, second position data, third position data, inertial measurement data, barometer data, meteorological data, radiation measurement data, and sonar data to obtain data correlation analysis results; The associated data are grouped according to the data association analysis results to obtain the corresponding relationship between the data optimization grouping and the data collected during the UAV flight.

[0010] Furthermore, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S102: According to the data optimization grouping, establish a list of data types that each data optimization grouping should contain; The data type list corresponding to each data optimization group is used in turn to identify the data collected during the flight of the UAV, and the data type identification result of the data collected during the flight of the UAV is obtained; The data type identification results of the data collected during the flight of the UAV are compared with the data optimization groups within the data optimization grouping to obtain the data integrity detection results of the data optimization grouping. If there is missing data in each data optimization group within the data optimization grouping in the data integrity detection results of the data optimization grouping, the missing data is determined in the data integrity detection results of the data optimization grouping, and the missing data is added to the corresponding data optimization grouping.

[0011] Furthermore, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S103: Constructing corresponding data correction models for the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping, and the seventh data optimization grouping, and storing the corresponding data correction models constructed for the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping, and the seventh data optimization grouping in a knowledge base, and setting data correction model matching conditions for the corresponding data correction models constructed for the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping, and the seventh data optimization grouping in the knowledge base; When the data in the data optimization group meets the matching conditions of the data correction model, the knowledge base retrieves the data correction model corresponding to the data optimization group; The first data optimization group corresponds to the first data correction model, the second data optimization group corresponds to the second data correction model, the third data optimization group corresponds to the third data correction model, the fourth data optimization group corresponds to the fourth data correction model, the fifth data optimization group corresponds to the fifth data correction model, the sixth data optimization group corresponds to the sixth data correction model, and the seventh data optimization group corresponds to the seventh data correction model.

[0012] Furthermore, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S103: a first data correction model, preprocessing night image video data and infrared thermal imaging data, extracting edge, texture and brightness features from the preprocessed night image video data to obtain night image video data features, extracting temperature distribution features from the infrared thermal imaging data to obtain infrared thermal imaging data features, matching the night image video data features with the infrared thermal imaging data features in time and space, eliminating feature deviations between the night image video data features and the infrared thermal imaging data features through geometric calibration, fusing the matched and calibrated night image video data with the infrared thermal imaging data to obtain a first data correction model, the first data correction model is used to detect anomalies in the night image video data and the infrared thermal imaging data, and to correct the night image video data and the infrared thermal imaging data; A second data correction model is constructed to extract and fuse features of the daytime image video data and the first spectral data to obtain a second data correction model, and the second data correction model is used to correct the daytime image video data and the first spectral data; A third data correction model is constructed by extracting and fusing features of the laser scanning data and the first position data to obtain a third data correction model, and the third data correction model is used to correct the laser scanning data and the first position data; The fourth data is optimized and grouped, and features are extracted and fused on the inertial measurement data and the barometer data to construct a fourth data correction model, where the fourth data correction model is used to correct the inertial measurement data and the barometer data.

[0013] Furthermore, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S103: The fifth data optimization grouping is to extract and fuse the features of the second location data and the meteorological data to construct a fifth data correction model, which is used to correct the second location data and the meteorological data. The sixth data is optimized and grouped, and features are extracted and fused on the radiation measurement data and the second spectrum data to construct a sixth data correction model, which is used to correct the stimulated radiation measurement data and the second spectrum data.

[0014] Furthermore, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S103: The seventh data is optimized and grouped, and features are extracted and fused on the sonar data and the third position data to construct a seventh data correction model, which is used to correct the sonar data and the third position data.

[0015] Furthermore, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S104: According to the data processing results of the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping and the seventh data optimization grouping, cluster analysis is performed to obtain risk assessment indicators, which include altitude error, flight attitude stability, weather condition suitability and obstacle distance. Risk assessment indicators are extracted for the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping and the seventh data optimization grouping to obtain altitude error characteristics, flight attitude stability characteristics, weather condition suitability characteristics and obstacle distance characteristics.

[0016] Furthermore, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S105: Acquire the real-time flight monitoring data of the UAV, parse the received real-time flight monitoring data of the UAV, and extract various types of data in the real-time flight monitoring data of the UAV, wherein the various types of data in the real-time flight monitoring data of the UAV include real-time image video data, real-time infrared thermal imaging data, real-time spectrum data, real-time laser scanning data, real-time inertial measurement data, real-time barometer data, real-time position data, and real-time meteorological data; Each type of data in the real-time flight monitoring data of the UAV is matched and stored in the corresponding data optimization group, and a real-time first data optimization group, a real-time second data optimization group, a real-time third data optimization group, a real-time fourth data optimization group, a real-time fifth data optimization group, a real-time sixth data optimization group, and a real-time seventh data optimization group are obtained; The correspondence between various types of data in the real-time flight monitoring data of the UAV and the real-time data optimization grouping is as follows: real-time night image video data and real-time infrared thermal imaging data are classified into the real-time first data optimization group, real-time daytime image video data and real-time first spectral data are classified into the real-time second data optimization group, real-time laser scanning data and real-time first position data are classified into the real-time third data optimization group, real-time inertial measurement data and real-time barometer data are classified into the real-time fourth data optimization group, real-time second position data and real-time meteorological data are classified into the real-time fifth data optimization group, real-time radiation measurement data and real-time second spectral data are classified into the real-time sixth data optimization group, and real-time sonar data and real-time third position data are classified into the real-time seventh data optimization group.

[0017] Beneficial effects of the present invention: The present invention significantly improves the accuracy and reliability of data collected during the flight of the UAV through data optimization grouping, feature extraction and fusion, and the construction of a special data correction model, which helps to reduce the impact of data errors and noise on risk assessment results. The random forest model is trained using historical UAV flight data and risk assessment results, and combined with multiple evaluation indicators such as altitude error, flight attitude stability, weather conditions suitability, and obstacle distance. The present invention can more comprehensively evaluate the flight risk of the UAV and provide more accurate risk assessment results. By acquiring and processing the flight monitoring data of the UAV in real time, the present invention can monitor the flight status of the UAV in real time, promptly discover and deal with potential flight risks, and thus avoid the occurrence of flight accidents. Based on the real-time risk assessment results, the present invention can provide scientific decision-making support for UAV operators, such as adjusting flight plans, changing flight altitudes or paths, taking obstacle avoidance measures, etc., to ensure the safe flight of the UAV.

[0018] By comprehensively applying a variety of technical means, the present invention effectively reduces various risks that drones may face during flight, providing a strong guarantee for the safe flight of drones. Through accurate risk assessment and real-time flight monitoring, the present invention helps to optimize the flight path and strategy of drones, reduce unnecessary flight time and energy consumption, and thus improve flight efficiency.

[0019] In summary, the beneficial effects of the present invention are not only reflected in improving the accuracy and real-time performance of risk assessment, but also in comprehensively improving the flight safety and efficiency of drones by enhancing data security and providing decision support. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0021] Figure 1 A schematic diagram of a method flow chart provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. The technical solutions provided by the embodiments of the present invention are described in detail below in conjunction with the drawings.

[0023] In order to better understand the purpose of the present invention, the present invention is described in further detail below.

[0024] See also Figure 1 The present invention provides a method for risk assessment and quality monitoring of unmanned aerial vehicle flight, comprising: The present invention provides a method for unmanned aerial vehicle flight risk assessment and quality monitoring, comprising: Step S101, acquiring data collected during the flight of the drone, the data collected during the flight of the drone including image and video data, infrared thermal imaging data, spectral data, laser scanning data, inertial measurement data, barometer data, position data, meteorological data, radiation measurement data and sonar data, the image and video data including night image and video data and day image and video data; Image and video data: including night image and video data and day image and video data, which are used for visual monitoring of the UAV's flight environment, target tracking, obstacle detection, etc.

[0025] Infrared thermal imaging data: Provides temperature distribution information on the surface of objects, which helps detect targets at night or in poor visibility, and monitors the temperature of the drone's power system.

[0026] Spectral data: used to analyze the chemical composition of the environment, vegetation health, water quality, etc., for scenarios such as environmental monitoring and agricultural applications.

[0027] Laser scanning data: 3D point cloud data generated by laser ranging, used to build an accurate 3D model of the surrounding environment, which helps with obstacle avoidance and path planning.

[0028] Inertial measurement data: including data provided by accelerometers and gyroscopes, which are used to monitor the attitude, angular velocity and acceleration of the drone and are the basis of flight control and navigation.

[0029] Barometer data: Provides current altitude information, which can be combined with GPS data to accurately determine the three-dimensional position of the drone.

[0030] Location data: obtained through GPS or other global navigation satellite systems (GNSS) and used to track the geographic location of the drone in real time.

[0031] Meteorological data: including wind speed, wind direction, temperature, humidity, etc., used to assess the impact of current weather conditions on flight and prevent flight risks caused by bad weather.

[0032] Radiation measurement data: Monitoring the level of ionizing radiation in the environment is crucial for scenarios such as nuclear facility monitoring and environmental protection.

[0033] Sonar data: Mainly used for underwater drones, it uses sound waves to measure distance and perceive the environment, which helps with underwater navigation and obstacle avoidance.

[0034] By acquiring diversified data, the method of the present invention can comprehensively evaluate the flight status, environmental adaptability, potential risks, etc. of the drone, thereby providing strong support for flight safety and quality monitoring. The data is collected in real time by various sensors of the drone, and after subsequent data processing, optimization, feature extraction and other steps, an accurate assessment of the flight risk of the drone is finally formed.

[0035] Step S102, establishing data optimization groups, the data optimization groups include a first data optimization group, a second data optimization group, a third data optimization group, a fourth data optimization group, a fifth data optimization group, a sixth data optimization group and a seventh data optimization group, copying the position data to generate first position data, second position data and third position data, copying the spectral data to generate first spectral data and second spectral data, classifying night image video data and infrared thermal imaging data into the first data optimization group, classifying day image video data and the first spectral data into the second data optimization group, classifying laser scanning data and the first position data into the third data optimization group, classifying inertial measurement data and barometer data into the fourth data optimization group, classifying the second position data and meteorological data into the fifth data optimization group, classifying radiation measurement data and the second spectral data into the sixth data optimization group, and classifying sonar data and the third position data into the seventh data optimization group; In step S102, the present invention proposes a refined data optimization grouping strategy, aiming to improve the processing efficiency and accuracy of UAV flight data through reasonable grouping. The following is a detailed data optimization grouping process: The data optimization grouping includes seven groups: the first data optimization grouping to the seventh data optimization grouping. Each grouping is divided according to different data types and optimization requirements, so as to perform independent and targeted data processing and correction later.

[0036] The first data optimization grouping: includes night image video data and infrared thermal imaging data.

[0037] Reasons for grouping: Night image video data and infrared thermal imaging data are particularly important at night or under poor visibility conditions. The two have certain complementary characteristics (such as images provide visual details and infrared provides temperature information), so they are classified into the same group for joint processing.

[0038] The second data optimization grouping includes daytime image video data and the first spectrum data.

[0039] Rationale for grouping: Daytime imagery video data provides rich visual information, while spectral data provides detailed information about the physical and chemical properties of the environment. Combining the two helps to more fully understand the flight environment.

[0040] The third data optimization grouping includes the laser scanning data and the first position data.

[0041] Reason for grouping: Laser scanning data provides the precise three-dimensional structure of the surrounding environment, while position data provides the spatial position of the drone. Combining the two helps achieve accurate obstacle avoidance and path planning.

[0042] The fourth data optimization grouping includes inertial measurement data and barometer data.

[0043] Reason for grouping: Inertial measurement data provides information about the drone's attitude, angular velocity, and acceleration, while barometer data provides altitude information. Both are core data for flight control and navigation, so they are grouped together for processing.

[0044] The fifth data optimization grouping includes the second location data and meteorological data.

[0045] Reason for grouping: Position data is used to determine the specific location of the drone, while weather data provides real-time weather conditions of the current flight environment. Combining the two helps to assess the impact of weather conditions on flight.

[0046] The sixth data optimization grouping includes the data radiation measurement data and the second spectrum data.

[0047] Reason for grouping: Radiometric data focuses on the radiation level in the environment, while spectral data provides more information about the composition of the environment. The two are highly correlated in application scenarios such as environmental monitoring.

[0048] The seventh data optimization group: includes sonar data and third position data.

[0049] Reason for grouping: Sonar data is mainly used for underwater drones to provide sonar images or distance information of the underwater environment, while position data is used to determine the underwater position of the drone. Combining the two helps to achieve underwater navigation and obstacle avoidance.

[0050] Location data replication: To optimize data processing and redundant backup, location data is replicated three times to generate first location data, second location data, and third location data. This can increase data availability and provide the possibility of recovery when some data is lost.

[0051] Spectral data replication: The spectral data is also replicated twice to generate the first spectral data and the second spectral data to meet the needs of different groups.

[0052] Through the data optimization grouping strategy, the present invention can independently and efficiently process different types of data, and the data in each group has similarity or complementarity, which is conducive to subsequent data correction, feature extraction and risk assessment. In addition, the replication and generation of data further enhances the flexibility and reliability of data processing.

[0053] Step S103, optimizing and correcting the data in the first data optimization group, the second data optimization group, the third data optimization group, the fourth data optimization group, the fifth data optimization group, the sixth data optimization group, and the seventh data optimization group in sequence, to obtain corrected data collected during the flight of the drone; In step S103, the technical solution of the present invention performs detailed optimization and correction processing on the data in each data optimization group to eliminate data errors and noise and improve the accuracy and reliability of the data. The following is the optimization and correction process for each data optimization group: First data optimization grouping (night image video data and infrared thermal imaging data); Optimization and correction content: Noise removal, color correction and video stabilization processing are performed on night image video data to improve image quality. Calibration and calibration of infrared thermal imaging data are performed to eliminate temperature measurement deviations, and noise suppression and image enhancement processing are performed.

[0054] Joint processing: Match night image video data with infrared thermal imaging data in time and space, eliminate feature deviations between the two through geometric calibration, and achieve data fusion and complementarity.

[0055] Second data optimization grouping (daytime image video data and first spectrum data); Optimization and correction content: Noise removal, image sharpening and color correction are performed on daytime image video data to enhance image clarity and authenticity. Spectral data is smoothed, spectral corrected and feature extracted to obtain accurate environmental information.

[0056] Joint processing: feature-level fusion of image and video data with spectral data, combining visual information and spectral characteristics to improve the overall understanding of the environment.

[0057] The third data optimization grouping (laser scanning data and first position data); Optimization and correction content: De-noise, point cloud registration and hole filling of laser scanning data to obtain a complete and accurate 3D environment model. Filter the first position data to reduce positioning errors.

[0058] Joint processing: The laser scanning data is integrated with the position data to achieve accurate correspondence between the three-dimensional environment and the actual position of the UAV, providing a basis for obstacle avoidance and path planning.

[0059] Fourth data optimization grouping (inertial measurement data and barometer data); Optimized correction content: Filter and integrate inertial measurement data to obtain more accurate attitude, angular velocity and acceleration information. Calibrate and temperature compensate barometer data to improve the accuracy of altitude measurement.

[0060] Joint processing: Inertial measurement data is integrated with barometer data to achieve accurate estimation of the drone’s attitude and position, providing key data support for flight control.

[0061] Fifth data optimization grouping (second location data and meteorological data); Optimization and correction content: Smooth the second position data to reduce the fluctuation of positioning data. Verify and interpolate the meteorological data to ensure the continuity and accuracy of meteorological information.

[0062] Joint processing: Fusion analysis of location data and meteorological data to assess the impact of current meteorological conditions on UAV flight safety and provide a basis for flight decision-making.

[0063] Sixth data optimization grouping (radiometric data and second spectral data); Optimization and correction content: Correct and normalize the radiation measurement data to eliminate measurement deviations. De-noise and feature extract the second spectrum data to obtain effective environmental information.

[0064] Joint processing: Combine radiometric data with spectral data to assess radiation levels and composition characteristics in the environment, providing support for applications such as environmental monitoring.

[0065] The seventh data optimization grouping (sonar data and third position data); Optimization and correction content: Echo cancellation, noise suppression and image reconstruction processing are performed on sonar data to obtain clear underwater sonar images. Filtering and verification processing is performed on the third position data to improve the accuracy of underwater positioning.

[0066] Joint processing: Sonar data and position data are integrated and analyzed to achieve accurate spatial positioning and environmental perception capabilities of underwater drones.

[0067] Through the above-mentioned optimization and correction process, the present invention can significantly improve the quality of data collected during the flight of the UAV, and provide an accurate and reliable data basis for subsequent risk assessment and quality monitoring.

[0068] Step S104, extracting data features from the data correction results of the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping, and the seventh data optimization grouping to obtain height error features, flight attitude stability features, meteorological condition suitability features, and obstacle distance features, performing feature fusion on the height error features, flight attitude stability features, meteorological condition suitability features, and obstacle distance features to obtain risk assessment indicator feature fusion data, using the data collected during historical UAV flights and historical risk assessment results as training sets, training and verifying the random forest model, using the height error, flight attitude stability, meteorological condition suitability, obstacle distance, etc. as input features of the random forest model, and obtaining a UAV flight risk assessment model; In step S104, the method of the present invention involves further processing the data of each optimized data group after optimization and correction to extract key features and construct a UAV flight risk assessment model. The following is a detailed explanation of this step: The first data optimization grouping: mainly extracts features related to the night flight environment, such as temperature distribution differences in infrared thermal imaging, edge detection and contrast features in night images, etc. These features can be used to evaluate the visibility and obstacle recognition capabilities of night flights.

[0069] Second data optimization grouping: extracting texture, color and shape features in daytime images, as well as specific wavelength absorption or reflection features in spectral data. These features reflect the physical and chemical properties of the environment and help assess the complexity and potential risks of the flight environment.

[0070] The third data optimization grouping: extract three-dimensional spatial structural features from the laser scanning data, such as obstacle distance, terrain undulation, etc., and combine the position data to evaluate the flight path and obstacle avoidance capability of the drone.

[0071] Fourth data optimization grouping: extract angular velocity, acceleration and attitude angle features from inertial measurement data, as well as altitude and air pressure change features from barometer data. These features are directly related to the flight stability and controllability of the drone.

[0072] Fifth data optimization grouping: extract environmental parameter characteristics such as temperature, humidity, wind speed, etc. from meteorological data, and combine with location data to evaluate the impact of meteorological conditions on flight safety.

[0073] The sixth data optimization group: Extracting radiation intensity and spectral features from radiation measurement and spectral data to assess the radiation level and composition in the environment is particularly important for specific applications (such as nuclear radiation monitoring).

[0074] Seventh data optimization group (if applicable): Extract underwater obstacle distance, seabed topography and other features from sonar data. For underwater drones, these features are directly related to their obstacle avoidance capabilities and path planning.

[0075] The extracted features include altitude error features, flight attitude stability features, meteorological condition suitability features, and obstacle distance features, etc. These features are then fused to form a comprehensive risk assessment indicator feature fusion data.

[0076] Next, the random forest model was trained and validated using the data collected during historical drone flights and historical risk assessment results as training sets. Random forest is an ensemble learning method that improves the accuracy of classification or regression by building multiple decision trees. In this step, features such as altitude error, flight attitude stability, weather condition suitability, and obstacle distance were used as inputs to the random forest model. Through repeated training and adjustment of model parameters, a model that can accurately assess drone flight risks was finally obtained.

[0077] Through this step, the present invention successfully integrates multi-source data features into risk assessment indicators, and uses machine learning technology to build a model that can assess the flight risks of drones in real time, providing strong protection for the safe flight of drones.

[0078] Step S105, obtaining the real-time flight monitoring data of the UAV, classifying the real-time flight monitoring data of the UAV according to the data characteristics, obtaining the classification results of the characteristics of the real-time flight monitoring data of the UAV, grouping the classification results of the characteristics of the real-time flight monitoring data of the UAV, and correcting the characteristics of the grouped real-time flight monitoring data of the UAV to obtain the corrected real-time flight monitoring data characteristics of the UAV, substituting the corrected real-time flight monitoring data characteristics of the UAV into the UAV flight risk assessment model to obtain the comprehensive flight risk assessment results of the UAV, which include the risk assessment results of the altitude error, the risk assessment results of the flight attitude stability, the risk assessment results of the weather conditions suitability and the risk assessment results of the obstacle distance.

[0079] In step S105, the method of the present invention describes in detail how to evaluate the flight risk of the drone in real time. The following is the detailed execution process of this step: Real-time acquisition of various data collected by the drone during flight, including but not limited to real-time image and video data, real-time infrared thermal imaging data, real-time spectral data, real-time laser scanning data, real-time inertial measurement data, real-time barometer data, real-time location data, and real-time meteorological data.

[0080] After acquiring real-time data, the data is classified according to its characteristics, and data with similar characteristics and uses are grouped together. Then, according to the classification results, the data is assigned to pre-defined data optimization groups, namely the first data optimization group to the seventh data optimization group. This step ensures the efficiency and accuracy of subsequent data processing.

[0081] Correct the real-time flight monitoring data in each group to eliminate potential errors and noise. The correction process may include multiple technical means such as denoising, filtering, calibration, and calibration. The specific method depends on the type and characteristics of the data. For example, noise removal and color correction are performed on real-time image and video data, temperature calibration is performed on real-time infrared thermal imaging data, and point cloud calibration is performed on real-time laser scanning data.

[0082] The corrected real-time flight monitoring data features are substituted into the trained UAV flight risk assessment model. This model is built based on random forest or other suitable machine learning algorithms, and can calculate the comprehensive flight risk assessment results of the UAV based on the input feature data.

[0083] The UAV flight risk assessment model calculates the comprehensive flight risk assessment results of the UAV based on the input real-time flight monitoring data characteristics. The assessment results include: Altitude error risk assessment results: Evaluate the degree of error between the drone and the predetermined altitude during flight to determine whether there is a risk of altitude deviation.

[0084] Flight attitude stability risk assessment results: Evaluate whether the UAV’s flight attitude is stable and whether there is a risk of attitude loss of control.

[0085] Weather conditions suitability risk assessment results: Assess the impact of current weather conditions on drone flight safety based on real-time meteorological data. Bad weather such as strong winds and heavy rains may increase flight risks.

[0086] Obstacle distance risk assessment results: Assess the distance between the drone and surrounding obstacles and their potential collision risk based on data such as laser scanning or sonar.

[0087] Based on the calculated comprehensive flight risk assessment results, the system can conduct real-time risk assessment of the drone's flight status. If a high-risk area or potential danger is found, the system will immediately warn the drone operator and provide corresponding decision support suggestions, such as adjusting the flight plan, changing the flight altitude or path, taking obstacle avoidance measures, etc., to ensure the drone's safe flight.

[0088] Through this step, the method of the present invention can realize real-time monitoring and evaluation of the flight risks of the UAV, providing strong technical support for the safe flight of the UAV.

[0089] Specifically, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S101: Establish data preprocessing batches according to the type and source of data collected during the drone flight, store the data collected during the drone flight in the blockchain in sequence according to the data preprocessing batches, and generate hash values ​​corresponding to the data collected during the drone flight; Preprocess the data collected during the flight of the drone, store the data generated during the preprocessing of the data collected during the flight of the drone and the data collected during the flight of the drone after the preprocessing is completed in the blockchain, and generate the corresponding hash value; When receiving the instruction to call the data information collected during the flight of the drone, recalculate the hash value corresponding to the data collected during the flight of the called drone, and match the calculation result with the original hash value of the data collected during the flight of the drone; If the hash values ​​match, the data collected during the flight of the called drone has not been tampered with. If the hash values ​​do not match, the data collected during the flight of the called drone has been tampered with.

[0090] Specifically, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S102: receiving a target of data optimization grouping, and determining a data type to be classified according to the target of data optimization grouping, wherein the data type to be classified includes night image video data, day image video data, infrared thermal imaging data, first spectrum data, second spectrum data, laser scanning data, first position data, second position data, third position data, inertial measurement data, barometer data, meteorological data, radiation measurement data, and sonar data; Perform correlation analysis on night image video data, day image video data, infrared thermal imaging data, first spectrum data, second spectrum data, laser scanning data, first position data, second position data, third position data, inertial measurement data, barometer data, meteorological data, radiation measurement data, and sonar data to obtain data correlation analysis results; The associated data are grouped according to the data association analysis results to obtain the corresponding relationship between the data optimization grouping and the data collected during the UAV flight.

[0091] Receive the goals of data optimization grouping and clarify the specific goals and requirements of data optimization grouping. The goals are based on different application scenarios, flight missions or data type characteristics.

[0092] According to the goal of data optimization grouping, the system determines the data types that need to be classified and processed, including but not limited to: Night image and video data: image and video collected at night or in low-light environments.

[0093] Daytime image and video data: image and video collected during the day.

[0094] Infrared thermal imaging data: temperature distribution data obtained through infrared thermal imaging technology.

[0095] Spectral data (first spectral data, second spectral data): data reflecting the specific wavelength characteristics of light reflected or emitted by the surface of an object.

[0096] Laser scanning data: three-dimensional spatial structure data obtained through laser scanning technology.

[0097] Position data (first position data, second position data, third position data): real-time position information of the drone.

[0098] Inertial measurement data: angular velocity, acceleration and other data obtained by the inertial measurement unit (IMU).

[0099] Barometer data: data reflecting the flight altitude and air pressure changes of the drone.

[0100] Meteorological data: including real-time meteorological information such as temperature, humidity, and wind speed.

[0101] Radiation measurement data: Data that measures the radiation levels in an environment.

[0102] Sonar data: For underwater drones, underwater environment data obtained through sonar technology.

[0103] Next, the system performs correlation analysis on the various data types mentioned above. Correlation analysis aims to identify the intrinsic connections and mutual influences between different types of data. The analysis involves statistical methods, machine learning algorithms, or domain expert knowledge to determine which data types are logically or physically more closely related.

[0104] Based on the results of correlation analysis, the system groups related data into different data optimization groups. The data types within each group are more consistent in logic or processing flow, which facilitates subsequent data optimization and feature extraction.

[0105] Finally, a clear correspondence between the data optimization grouping and the data actually collected during the UAV flight is established. This correspondence ensures that the relevant data can be accurately called and processed in the subsequent data processing and analysis process.

[0106] Through the above steps, the method of the present invention can efficiently classify and optimize the grouping processing of large amounts of diverse data collected during the flight of the UAV, laying a solid foundation for subsequent data correction, feature extraction and risk assessment.

[0107] Specifically, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S102: According to the data optimization grouping, establish a list of data types that each data optimization grouping should contain; The data type list corresponding to each data optimization group is used in turn to identify the data collected during the flight of the UAV, and the data type identification result of the data collected during the flight of the UAV is obtained; The data type identification results of the data collected during the flight of the UAV are compared with the data optimization groups within the data optimization grouping to obtain the data integrity detection results of the data optimization grouping. If there is missing data in each data optimization group within the data optimization grouping in the data integrity detection results of the data optimization grouping, the missing data is determined in the data integrity detection results of the data optimization grouping, and the missing data is added to the corresponding data optimization grouping.

[0108] In step S102, the UAV flight risk assessment and quality monitoring method of the present invention further refines the process of data optimization grouping to ensure that the data type in each group is complete and accurate. The following is the specific execution process of this step: First, based on the determined data optimization groups (such as the first data optimization group, the second data optimization group, etc.), a list of data types that should be included in each group is established. These lists clearly specify the data types that need to be collected and processed in each group.

[0109] Next, the system uses the data type list corresponding to each data optimization group to identify the data actually collected during the drone flight. The purpose of this step is to match the real-time collected data with the pre-defined data types to determine which group each data belongs to.

[0110] After the data type identification is completed, the system compares the identification results with the predefined data type list in each data optimization group. This comparison process is to check whether the actual collected data type is consistent with the grouping requirements, so as to find any potential data missing or errors.

[0111] By comparing, the system can generate data integrity test results for data optimization groups. If the test results show that there is data missing in a group (that is, the actual data type collected does not match the grouping requirements), the system will further analyze to determine the specific missing data type.

[0112] When missing data types are identified, the system will take appropriate measures to complete the data based on the actual situation. This includes restoring data from backup data sources, generating missing data through data interpolation or predictive models, or simply marking the missing data for subsequent analysis. In most cases, if the data is missing so seriously that it affects the accuracy of the risk assessment, the system may trigger a warning or take other corrective measures.

[0113] Finally, any completed or updated data is added to the corresponding data optimization group so that each group contains a complete and accurate data set. In this way, subsequent data correction, feature extraction, and risk assessment steps can be performed based on high-quality data, thereby improving the accuracy and reliability of the entire assessment process.

[0114] Through the above steps, the method of the present invention ensures that the data collected during the flight of the UAV can be effectively classified, optimized and grouped, providing a solid foundation for subsequent risk assessment and quality monitoring.

[0115] Specifically, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S103: Constructing corresponding data correction models for the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping, and the seventh data optimization grouping, and storing the corresponding data correction models constructed for the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping, and the seventh data optimization grouping in a knowledge base, and setting data correction model matching conditions for the corresponding data correction models constructed for the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping, and the seventh data optimization grouping in the knowledge base; When the data in the data optimization group meets the matching conditions of the data correction model, the knowledge base retrieves the data correction model corresponding to the data optimization group; The first data optimization group corresponds to the first data correction model, the second data optimization group corresponds to the second data correction model, the third data optimization group corresponds to the third data correction model, the fourth data optimization group corresponds to the fourth data correction model, the fifth data optimization group corresponds to the fifth data correction model, the sixth data optimization group corresponds to the sixth data correction model, and the seventh data optimization group corresponds to the seventh data correction model.

[0116] Specifically, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S103: a first data correction model, preprocessing night image video data and infrared thermal imaging data, extracting edge, texture and brightness features from the preprocessed night image video data to obtain night image video data features, extracting temperature distribution features from the infrared thermal imaging data to obtain infrared thermal imaging data features, matching the night image video data features with the infrared thermal imaging data features in time and space, eliminating feature deviations between the night image video data features and the infrared thermal imaging data features through geometric calibration, fusing the matched and calibrated night image video data with the infrared thermal imaging data to obtain a first data correction model, the first data correction model is used to detect anomalies in the night image video data and the infrared thermal imaging data, and to correct the night image video data and the infrared thermal imaging data; A second data correction model is constructed to extract and fuse features of the daytime image video data and the first spectral data to obtain a second data correction model, and the second data correction model is used to correct the daytime image video data and the first spectral data; A third data correction model is constructed by extracting and fusing features of the laser scanning data and the first position data to obtain a third data correction model, and the third data correction model is used to correct the laser scanning data and the first position data; The fourth data is optimized and grouped, and features are extracted and fused on the inertial measurement data and the barometer data to construct a fourth data correction model, where the fourth data correction model is used to correct the inertial measurement data and the barometer data.

[0117] Specifically, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S103: The fifth data optimization grouping is to extract and fuse the features of the second location data and the meteorological data to construct a fifth data correction model, which is used to correct the second location data and the meteorological data. The sixth data is optimized and grouped, and features are extracted and fused on the radiation measurement data and the second spectrum data to construct a sixth data correction model, which is used to correct the stimulated radiation measurement data and the second spectrum data.

[0118] Specifically, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S103: The seventh data is optimized and grouped, and features are extracted and fused on the sonar data and the third position data to construct a seventh data correction model, which is used to correct the sonar data and the third position data.

[0119] In step S103, the UAV flight risk assessment and quality monitoring method of the present invention optimizes the grouping of different types of data and constructs multiple special data correction models to ensure the accuracy and reliability of various types of data. The following is a detailed description of each data correction model: First data correction model; Data processing object: night image video data and infrared thermal imaging data Processing flow: Preprocess night image video data and infrared thermal imaging data to improve data quality.

[0120] Edge, texture and brightness features are extracted from the preprocessed night image video data to obtain night image video data features.

[0121] The temperature distribution characteristics are extracted from the infrared thermal imaging data to obtain the infrared thermal imaging data characteristics.

[0122] Match the night image video data features with the infrared thermal imaging data features in time and space to ensure synchronization between the two.

[0123] The characteristic deviation between the night image video data features and the infrared thermal imaging data features is eliminated through geometric calibration.

[0124] The matched and calibrated night image video data is fused with the infrared thermal imaging data to generate fused data.

[0125] The first data correction model is used to detect whether there are anomalies in the fused data and to correct the abnormal data.

[0126] Second data correction model; Data processing object: daytime image video data and first spectrum data Processing flow: Feature extraction is performed on the daytime image video data and the first spectrum data to obtain image features and spectrum features respectively.

[0127] The image features are fused with the spectral features to construct a second data correction model.

[0128] The model is used to correct the daytime image video data and the first spectrum data to improve the accuracy and consistency of the data.

[0129] The third data correction model; Data processing object: laser scanning data and first position data Processing flow: Feature extraction is performed on the laser scanning data and the first position data to obtain three-dimensional spatial structure features and position features respectively.

[0130] The two features are combined to construct a third data correction model.

[0131] The model is used to correct the laser scanning data and the first position data to ensure the accuracy of the spatial position and the laser scanning data.

[0132] Fourth data correction model; Data processing object: inertial measurement data and barometer data Processing flow: Feature extraction is performed on the inertial measurement data and barometer data to obtain features such as angular velocity, acceleration and air pressure.

[0133] These features are integrated to construct the fourth data correction model.

[0134] This model is used to correct inertial measurement data and barometer data to improve the accuracy of flight attitude and altitude data.

[0135] Fifth data correction model; Data processing object: Second location data and meteorological data Processing flow: Feature extraction is performed on the second location data and the meteorological data to obtain location features and meteorological features (such as temperature, humidity, wind speed, etc.) respectively.

[0136] The two features are combined to construct the fifth data correction model.

[0137] The model is used to correct the second position data and the meteorological data to ensure the matching of the position data with the real-time meteorological conditions.

[0138] Sixth data correction model; Data processing object: radiation measurement data and second spectrum data Processing flow: Feature extraction is performed on the radiation measurement data and the second spectrum data to obtain radiation features and spectrum features respectively.

[0139] The two features are combined to construct the sixth data correction model.

[0140] The model is used to correct the radiation measurement data and the second spectrum data to improve the accuracy of the radiation level measurement.

[0141] Seventh data correction model; Data processing objects: sonar data and third-party location data (for underwater drones) Processing flow: Feature extraction is performed on the sonar data and the third position data to obtain underwater environment features and position features respectively.

[0142] The two features are combined to construct the seventh data correction model.

[0143] This model is used to correct sonar data and third-party location data to ensure the accuracy of underwater environment data and UAV location information.

[0144] By constructing the above-mentioned data correction model, the method of the present invention can comprehensively and carefully process and correct various types of data collected during the flight of the UAV, thereby improving the accuracy and reliability of the UAV flight risk assessment and quality monitoring.

[0145] Specifically, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S104: According to the data processing results of the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping and the seventh data optimization grouping, cluster analysis is performed to obtain risk assessment indicators, which include altitude error, flight attitude stability, weather condition suitability and obstacle distance. Risk assessment indicators are extracted for the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping and the seventh data optimization grouping to obtain altitude error characteristics, flight attitude stability characteristics, weather condition suitability characteristics and obstacle distance characteristics.

[0146] In step S104, the UAV flight risk assessment and quality monitoring method of the present invention further performs cluster analysis and extracts key risk assessment indicators by deeply processing the results of each data optimization grouping. The following is the specific content of this step: All data processing results from the first data optimization grouping to the seventh data optimization grouping are summarized. These results have been optimized and corrected to ensure the accuracy and reliability of the data.

[0147] The aggregated data are processed using cluster analysis techniques, with the goal of classifying the data according to similarities and associations, thereby identifying key risk factors in drone flight.

[0148] Based on the cluster analysis, the following four main risk assessment indicators are further extracted: Height Error: Feature extraction: By analyzing location data (such as GPS location data, barometer data, etc.), the difference between the actual flight altitude of the drone and the predetermined flight altitude is calculated.

[0149] Purpose: To evaluate the ability of UAVs to maintain a predetermined flight altitude and the impact that altitude changes may have on flight safety.

[0150] Flight attitude stability: Feature extraction: Use inertial measurement data (such as accelerometer, gyroscope data) and barometer data to analyze the changes in the drone's attitude angle and angular velocity.

[0151] Purpose: To evaluate the attitude stability of the UAV during flight, including the stability of roll, pitch and yaw, to ensure smooth flight.

[0152] Suitability of meteorological conditions: Feature extraction: Combine meteorological data (such as temperature, humidity, wind speed, wind direction, etc.) and location data to evaluate the impact of current meteorological conditions on drone flight.

[0153] Purpose: To identify adverse weather conditions that may have an adverse effect on flight, such as strong winds, heavy rain, thunderstorms, etc., to ensure flight safety.

[0154] Obstacle distance: Feature extraction: Using laser scanning data, image and video data (combined with image processing technology to identify obstacles) and sonar data (for underwater drones), the distance between the drone and surrounding obstacles is calculated.

[0155] Purpose: To assess the collision risk between the drone and potential obstacles, as well as the obstacle avoidance capability, to ensure flight safety.

[0156] Feature fusion: The extracted altitude error features, flight attitude stability features, meteorological condition suitability features and obstacle distance features are fused at the feature level to generate risk assessment indicator feature fusion data.

[0157] Model training and validation: The random forest model is trained and validated using the data collected during historical drone flights and historical risk assessment results as training sets. The above four risk assessment indicators are used as input features of the model, and the model parameters are optimized through multiple iterations to finally obtain a reliable drone flight risk assessment model.

[0158] Through the above steps, the method of the present invention can comprehensively and accurately evaluate the potential risks of the UAV during flight, provide powerful decision support for the UAV operator, and ensure flight safety.

[0159] Specifically, the UAV flight risk assessment and quality monitoring method of the present invention includes, in step S105: Acquire the real-time flight monitoring data of the UAV, parse the received real-time flight monitoring data of the UAV, and extract various types of data in the real-time flight monitoring data of the UAV, wherein the various types of data in the real-time flight monitoring data of the UAV include real-time image video data, real-time infrared thermal imaging data, real-time spectrum data, real-time laser scanning data, real-time inertial measurement data, real-time barometer data, real-time position data, and real-time meteorological data; Each type of data in the real-time flight monitoring data of the UAV is matched and stored in the corresponding data optimization group, and a real-time first data optimization group, a real-time second data optimization group, a real-time third data optimization group, a real-time fourth data optimization group, a real-time fifth data optimization group, a real-time sixth data optimization group, and a real-time seventh data optimization group are obtained; The correspondence between various types of data in the real-time flight monitoring data of the UAV and the real-time data optimization grouping is as follows: real-time night image video data and real-time infrared thermal imaging data are classified into the real-time first data optimization group, real-time daytime image video data and real-time first spectral data are classified into the real-time second data optimization group, real-time laser scanning data and real-time first position data are classified into the real-time third data optimization group, real-time inertial measurement data and real-time barometer data are classified into the real-time fourth data optimization group, real-time second position data and real-time meteorological data are classified into the real-time fifth data optimization group, real-time radiation measurement data and real-time second spectral data are classified into the real-time sixth data optimization group, and real-time sonar data and real-time third position data are classified into the real-time seventh data optimization group.

[0160] The classified data are allocated to the corresponding data optimization groups according to their characteristics and requirements. The characteristics of the grouped UAV real-time flight monitoring data are corrected, and the corresponding correction algorithm model in the knowledge base is applied to process the data of each data optimization group.

[0161] The technical solution of the present invention aims at the error problem of sensor data collected during the flight of UAVs in the air, and the noise data problem that may be introduced during the transmission of sensor data collected, and adopts a series of measures to improve the accuracy and reliability of data, thereby improving the accuracy of UAV flight risk assessment and quality monitoring. The following are specific solutions: Data optimization grouping and correction, establish data optimization grouping: group various data collected during the flight of the drone (such as image video data, infrared thermal imaging data, spectral data, etc.) according to type and optimization requirements. Through reasonable grouping, special optimization strategies can be formulated for different data types and application scenarios.

[0162] Data optimization and correction, optimize and correct the grouped data one by one, and by applying a special correction model or algorithm to the data in each group, the data error and noise interference can be significantly reduced.

[0163] Data feature extraction and fusion, after data optimization and correction, key features (such as altitude error features, flight attitude stability features, meteorological condition suitability features, obstacle distance features, etc.) are extracted from the corrected data through feature extraction methods.

[0164] These features are fused at the feature level to obtain more comprehensive and accurate risk assessment indicator feature fusion data. This step enhances the comprehensive expression ability of the data and provides a more reliable basis for subsequent risk assessment.

[0165] Risk assessment model construction and validation,The data collected during historical UAV flights and historical risk assessment results,are used as training sets to train and validate the random forest model.,The random forest model is used here because it can handle high-dimensional,data, is robust and not prone to overfitting.

[0166] The features such as altitude error, flight attitude stability, weather conditions suitability, obstacle distance, etc. are used as inputs to the random forest model to train the UAV flight risk assessment model. This model can output the comprehensive flight risk assessment results of the UAV based on the input feature data.

[0167] Real-time flight monitoring and risk assessment, during the real-time flight of the UAV, the real-time flight monitoring data is obtained, and grouped and corrected according to the previous data feature classification and grouping method. The corrected real-time flight monitoring data features are substituted into the trained UAV flight risk assessment model to obtain the real-time comprehensive flight risk assessment results of the UAV.

[0168] Based on the risk assessment results, decision support is provided to drone operators, such as adjusting flight plans, changing flight altitude or path, taking obstacle avoidance measures, etc.

[0169] In the data collection and preprocessing stage, the data is stored in batches on the blockchain and the corresponding hash values ​​are generated to ensure the security and integrity of the data. When the data needs to be called, the hash value is matched to verify whether the data has been tampered with.

[0170] To sum up, the technical solution of the present invention effectively solves the error and noise problems in sensor data collection and transmission during UAV flight through multiple links such as data optimization grouping, feature extraction and fusion, risk assessment model construction and verification, real-time flight monitoring and risk assessment, and significantly improves the accuracy of UAV flight risk assessment and quality monitoring.

[0171] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations, and the above-described embodiments of the present invention do not constitute a limitation on the scope of protection of the present invention.

Claims

1. A method for risk assessment and quality monitoring of unmanned aerial vehicle flight, characterized in that: include: Step S101, acquiring data collected during the flight of the drone, the data collected during the flight of the drone including image and video data, infrared thermal imaging data, spectral data, laser scanning data, inertial measurement data, barometer data, position data, meteorological data, radiation measurement data and sonar data, the image and video data including night image and video data and day image and video data; Step S102, establishing data optimization groups, the data optimization groups include a first data optimization group, a second data optimization group, a third data optimization group, a fourth data optimization group, a fifth data optimization group, a sixth data optimization group and a seventh data optimization group, copying the position data to generate first position data, second position data and third position data, copying the spectral data to generate first spectral data and second spectral data, classifying night image video data and infrared thermal imaging data into the first data optimization group, classifying day image video data and the first spectral data into the second data optimization group, classifying laser scanning data and the first position data into the third data optimization group, classifying inertial measurement data and barometer data into the fourth data optimization group, classifying the second position data and meteorological data into the fifth data optimization group, classifying radiation measurement data and the second spectral data into the sixth data optimization group, and classifying sonar data and the third position data into the seventh data optimization group; Step S103, optimizing and correcting the data in the first data optimization group, the second data optimization group, the third data optimization group, the fourth data optimization group, the fifth data optimization group, the sixth data optimization group, and the seventh data optimization group in sequence, to obtain corrected data collected during the flight of the drone; Step S104, extracting data features from the data correction results of the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping, and the seventh data optimization grouping to obtain height error features, flight attitude stability features, meteorological condition suitability features, and obstacle distance features, performing feature fusion on the height error features, flight attitude stability features, meteorological condition suitability features, and obstacle distance features to obtain risk assessment indicator feature fusion data, using the data collected during historical UAV flights and historical risk assessment results as training sets, training and verifying the random forest model, using the height error, flight attitude stability, meteorological condition suitability, and obstacle distance as input features of the random forest model to obtain a UAV flight risk assessment model; Step S105, obtaining the real-time flight monitoring data of the UAV, classifying the real-time flight monitoring data of the UAV according to the data characteristics, obtaining the classification results of the characteristics of the real-time flight monitoring data of the UAV, grouping the classification results of the characteristics of the real-time flight monitoring data of the UAV, and correcting the characteristics of the grouped real-time flight monitoring data of the UAV to obtain the corrected real-time flight monitoring data characteristics of the UAV, substituting the corrected real-time flight monitoring data characteristics of the UAV into the UAV flight risk assessment model to obtain the comprehensive flight risk assessment results of the UAV, which include the risk assessment results of the altitude error, the risk assessment results of the flight attitude stability, the risk assessment results of the weather conditions suitability and the risk assessment results of the obstacle distance.

2. The UAV flight risk assessment and quality monitoring method according to claim 1, characterized in that: In step S101, it includes: Establish data preprocessing batches according to the type and source of data collected during the drone flight, store the data collected during the drone flight in the blockchain in sequence according to the data preprocessing batches, and generate hash values ​​corresponding to the data collected during the drone flight; Preprocess the data collected during the flight of the drone, store the data generated during the preprocessing of the data collected during the flight of the drone and the data collected during the flight of the drone after the preprocessing is completed in the blockchain, and generate the corresponding hash value; When receiving the instruction to call the data information collected during the flight of the drone, recalculate the hash value corresponding to the data collected during the flight of the called drone, and match the calculation result with the original hash value of the data collected during the flight of the drone; If the hash values ​​match, the data collected during the flight of the called drone has not been tampered with. If the hash values ​​do not match, the data collected during the flight of the called drone has been tampered with.

3. The UAV flight risk assessment and quality monitoring method according to claim 1, characterized in that: In step S102, it includes: receiving a target of data optimization grouping, and determining a data type to be classified according to the target of data optimization grouping, wherein the data type to be classified includes night image video data, day image video data, infrared thermal imaging data, first spectrum data, second spectrum data, laser scanning data, first position data, second position data, third position data, inertial measurement data, barometer data, meteorological data, radiation measurement data, and sonar data; Perform correlation analysis on night image video data, day image video data, infrared thermal imaging data, first spectrum data, second spectrum data, laser scanning data, first position data, second position data, third position data, inertial measurement data, barometer data, meteorological data, radiation measurement data, and sonar data to obtain data correlation analysis results; The associated data are grouped according to the data association analysis results to obtain the corresponding relationship between the data optimization grouping and the data collected during the UAV flight.

4. The UAV flight risk assessment and quality monitoring method according to claim 3, characterized in that: In step S102, it includes: According to the data optimization grouping, establish a list of data types that each data optimization grouping should contain; The data type list corresponding to each data optimization group is used in turn to identify the data collected during the flight of the UAV, and the data type identification result of the data collected during the flight of the UAV is obtained; The data type identification results of the data collected during the flight of the UAV are compared with the data optimization groups within the data optimization grouping to obtain the data integrity detection results of the data optimization grouping. If there is missing data in each data optimization group within the data optimization grouping in the data integrity detection results of the data optimization grouping, the missing data is determined in the data integrity detection results of the data optimization grouping, and the missing data is added to the corresponding data optimization grouping.

5. The UAV flight risk assessment and quality monitoring method according to claim 1, characterized in that: In step S103, it includes: Constructing corresponding data correction models for the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping, and the seventh data optimization grouping, and storing the corresponding data correction models constructed for the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping, and the seventh data optimization grouping in a knowledge base, and setting data correction model matching conditions for the corresponding data correction models constructed for the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping, and the seventh data optimization grouping in the knowledge base; When the data in the data optimization group meets the matching conditions of the data correction model, the knowledge base retrieves the data correction model corresponding to the data optimization group; The first data optimization group corresponds to the first data correction model, the second data optimization group corresponds to the second data correction model, the third data optimization group corresponds to the third data correction model, the fourth data optimization group corresponds to the fourth data correction model, the fifth data optimization group corresponds to the fifth data correction model, the sixth data optimization group corresponds to the sixth data correction model, and the seventh data optimization group corresponds to the seventh data correction model.

6. The UAV flight risk assessment and quality monitoring method according to claim 5, characterized in that: In step S103, it includes: a first data correction model, preprocessing night image video data and infrared thermal imaging data, extracting edge, texture and brightness features from the preprocessed night image video data to obtain night image video data features, extracting temperature distribution features from the infrared thermal imaging data to obtain infrared thermal imaging data features, matching the night image video data features with the infrared thermal imaging data features in time and space, eliminating feature deviations between the night image video data features and the infrared thermal imaging data features through geometric calibration, fusing the matched and calibrated night image video data with the infrared thermal imaging data to obtain a first data correction model, the first data correction model is used to detect anomalies in the night image video data and the infrared thermal imaging data, and to correct the night image video data and the infrared thermal imaging data; A second data correction model is constructed to extract and fuse features of the daytime image video data and the first spectral data to obtain a second data correction model, and the second data correction model is used to correct the daytime image video data and the first spectral data; A third data correction model is constructed by extracting and fusing features of the laser scanning data and the first position data to obtain a third data correction model, and the third data correction model is used to correct the laser scanning data and the first position data; The fourth data is optimized and grouped, and features are extracted and fused on the inertial measurement data and the barometer data to construct a fourth data correction model, where the fourth data correction model is used to correct the inertial measurement data and the barometer data.

7. The UAV flight risk assessment and quality monitoring method according to claim 6, characterized in that: In step S103, it includes: The fifth data optimization grouping is to extract and fuse the features of the second location data and the meteorological data to construct a fifth data correction model, which is used to correct the second location data and the meteorological data. The sixth data is optimized and grouped, and features are extracted and fused on the radiation measurement data and the second spectrum data to construct a sixth data correction model, which is used to correct the stimulated radiation measurement data and the second spectrum data.

8. The UAV flight risk assessment and quality monitoring method according to claim 7, characterized in that: In step S103, it includes: The seventh data is optimized and grouped, and features are extracted and fused on the sonar data and the third position data to construct a seventh data correction model, which is used to correct the sonar data and the third position data.

9. The UAV flight risk assessment and quality monitoring method according to claim 1, characterized in that: In step S104, it includes: According to the data processing results of the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping and the seventh data optimization grouping, cluster analysis is performed to obtain risk assessment indicators, which include altitude error, flight attitude stability, weather condition suitability and obstacle distance. Risk assessment indicators are extracted for the first data optimization grouping, the second data optimization grouping, the third data optimization grouping, the fourth data optimization grouping, the fifth data optimization grouping, the sixth data optimization grouping and the seventh data optimization grouping to obtain altitude error characteristics, flight attitude stability characteristics, weather condition suitability characteristics and obstacle distance characteristics.

10. The UAV flight risk assessment and quality monitoring method according to claim 1, characterized in that: In step S105, it includes: Acquire the real-time flight monitoring data of the UAV, parse the received real-time flight monitoring data of the UAV, and extract various types of data in the real-time flight monitoring data of the UAV, wherein the various types of data in the real-time flight monitoring data of the UAV include real-time image video data, real-time infrared thermal imaging data, real-time spectrum data, real-time laser scanning data, real-time inertial measurement data, real-time barometer data, real-time position data, and real-time meteorological data; Each type of data in the real-time flight monitoring data of the UAV is matched and stored in the corresponding data optimization group, and a real-time first data optimization group, a real-time second data optimization group, a real-time third data optimization group, a real-time fourth data optimization group, a real-time fifth data optimization group, a real-time sixth data optimization group, and a real-time seventh data optimization group are obtained; The correspondence between various types of data in the real-time flight monitoring data of the UAV and the real-time data optimization grouping is as follows: real-time night image video data and real-time infrared thermal imaging data are classified into the real-time first data optimization group, real-time daytime image video data and real-time first spectral data are classified into the real-time second data optimization group, real-time laser scanning data and real-time first position data are classified into the real-time third data optimization group, real-time inertial measurement data and real-time barometer data are classified into the real-time fourth data optimization group, real-time second position data and real-time meteorological data are classified into the real-time fifth data optimization group, real-time radiation measurement data and real-time second spectral data are classified into the real-time sixth data optimization group, and real-time sonar data and real-time third position data are classified into the real-time seventh data optimization group.

Images