Low-altitude economic flight data management method and system based on edge calculation

Abstract

The invention relates to the technical field of data management, in particular to a low-altitude economic flight data management method and system based on edge computing, and the system comprises edge computing nodes deployed on an aircraft and a ground base station, a sensor for collecting multi-source data, a module for preprocessing the data, and an annular data queue caching mechanism. The system comprises a composite biological feature code generation module, a distributed storage edge node, an abnormal fluctuation feature extraction module, an equipment health state and trend analysis and calculation module, a virtual power supply manager, a main power supply, a standby power supply, an abnormal behavior monitoring module and a defense strategy library. After preprocessing and caching, feature codes are generated and encrypted distributed storage is carried out, dynamic resource adjustment is realized through state evaluation and a digital twinning and optimization algorithm, a neural network is utilized to manage a power supply, data security is guaranteed in combination with abnormal monitoring, and a whole-process management system is formed.

Description

Technical Field

[0001] The present invention relates to the technical field of data management, and particularly to a method and system for managing low-altitude economic flight data based on edge computing. Background Art

[0002] As an emerging economic form relying on the development of low-altitude airspace, the low-altitude economy shows great potential in scenarios such as unmanned aerial vehicle (UAV) logistics distribution, agricultural plant protection, and emergency rescue, and the degree of airspace opening and equipment density continue to increase.

[0003] However, its large-scale development faces double challenges of data management efficiency and security: in the traditional cloud processing mode, since data needs to be transmitted to a remote server over a long distance, the round-trip delay exceeds 200 ms, making it difficult to meet the real-time response requirements of low-altitude flight devices for sudden environmental changes (such as avoiding obstacles within 50 ms). At the same time, the computing resources of low-altitude flight devices are limited, and centralized data storage is easily targeted by attacks. The number of malicious intrusion incidents on the global low-altitude communication link increases by 67% annually, and the risk of data leakage is high. Moreover, different low-altitude tasks have large differences in the demand for computing power and communication resources. The traditional static resource allocation strategy results in a computing power utilization rate of only 55% for devices and a waste of communication bandwidth of more than 40%. The existing monitoring relies on threshold alarms and can only identify obvious faults. 73% of the hidden faults that cause accidents fail to give early warnings 24 hours in advance. The traditional data management mode has become a bottleneck for the large-scale development of the low-altitude economy, and there is an urgent need for a new data management method that integrates edge computing, intelligent algorithms, and active security mechanisms to achieve low-latency processing, dynamic resource optimization, and full-link risk prevention and control.

[0004] Based on the above problems, the present invention provides a method and system for managing low-altitude economic flight data based on edge computing. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention provides a method and system for managing low-altitude economic flight data based on edge computing, which solves problems such as significant real-time defects, weak data security systems, low resource scheduling efficiency, and insufficient fault prediction capabilities, and realizes a data management system with high real-time performance, high security, and high reliability in low-altitude economic scenarios.

[0006] To achieve the above object, one aspect of the present invention provides a method for managing low-altitude economic flight data based on edge computing, including: Collecting multi-source data of low-altitude flight devices through edge computing nodes deployed on aircraft and ground base stations, performing preprocessing operations on the data in the edge computing nodes, and at the same time using a circular data queue caching mechanism to segment and manage the data, configuring dynamically allocated cache space for each data segment, and performing data storage or cleaning operations according to the occupancy of the cache space; Generate a composite biometric code for the flight data in the cache, encrypt the data blocks using a chaotic encryption algorithm in a sharded manner, and store the encrypted data blocks distributively on multiple edge nodes. Verify whether the composite biometric code is consistent with the stored feature code on the edge computing node. If not, start a secondary verification mechanism to further confirm the user's identity; Extract the abnormal fluctuation characteristics of the flight data, calculate the device health status coefficient and the trend analysis coefficient based on the preprocessed data. At the same time, predict the future state of the aircraft by combining historical flight data and multi-dimensional parameters collected in real time. If it exceeds the threshold, trigger a fault warning; Build a digital twin model based on the current flight data and the resource allocation scheme, and predict the flight trajectory and energy consumption through a physical simulation engine. If the deviation between the simulation result and the preset threshold exceeds 5%, trigger an adaptive entropy weight hybrid optimization algorithm, and generate a priority queue according to the flight task urgency and resource weights to achieve dynamic adjustment of the resource allocation strategy and iteratively adjust the resource allocation parameters; Input the power consumption characteristic data of the aircraft into a heterogeneous device management neural network for power management prediction, generate device power consumption prediction parameters, and perform power response control on the main power supply and standby power supply of the aircraft through a virtual power manager to generate power adjustment parameters; Monitor the user's operation behavior on the aircraft data. When an abnormal behavior is detected, record the behavior and store it in the defense policy library for subsequent security policy optimization. At the same time, use the real-time monitoring ability of the edge computing node to quickly respond to abnormal behaviors to ensure the security and reliability of the aircraft data.

[0007] The second aspect of the present invention is to provide a low-altitude economy flight data management system based on edge computing for implementing the low-altitude economy flight data management method based on edge computing, including the following modules: Edge data collection and preprocessing module: Through the edge computing nodes deployed on the aircraft and the ground base station, collect multi-source data such as device status, communication signals, and environmental parameters in real time, perform denoising and standardization preprocessing, and adopt a circular data queue caching mechanism to manage the data in segments according to time windows, and dynamically clean redundant data according to the cache occupancy; Distributed security storage and verification module: Generate a composite biometric code containing the device hardware fingerprint and the user operation characteristics, encrypt the data blocks in a sharded manner by combining a dynamic chaotic encryption algorithm, and store them in multiple edge nodes through a distributed hash table; When accessing, first verify the feature code, and trigger a short message verification code or biometric secondary verification in case of an anomaly; Intelligent status evaluation and prediction module: Calculate the device health status coefficient and the trend analysis coefficient based on multi-dimensional data, detect abnormal fluctuations through the Z-score and DBSCAN algorithms, generate sub-fluctuation impact factors, trigger a fault warning when exceeding the threshold, and output the fault type and the remaining time; Digital Twin-driven Resource Optimization Module: Build a lightweight digital twin model to simulate flight trajectories and energy consumption. If the deviation exceeds 5%, trigger the adaptive entropy weight algorithm to dynamically allocate computing power and bandwidth resources according to the task urgency and generate a priority queue. Intelligent Power Management Module: Use the LSTM neural network to predict power consumption parameters, coordinate the main power supply and supercapacitors through a virtual power manager, execute a stepped power supply strategy, and optimize power distribution in combination with the Bayesian network to extend the battery life. Abnormal Behavior Monitoring Module: Establish a user operation baseline through unsupervised learning, real-time monitor indicators such as access frequency and path depth, block the connection when an anomaly is detected, record the characteristics in the defense strategy library, and update the security model.

[0008] A third aspect of the present invention is to provide a computer device, which includes: a processor and a memory. At least one instruction, at least one program, a code set, or an instruction set is stored in the memory, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement a method for managing low-altitude economic flight data based on edge computing.

[0009] A fourth aspect of the present invention is to provide a computer-readable storage medium, in which at least one instruction, at least one program, a code set, or an instruction set is stored, and the at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement a method for managing low-altitude economic flight data based on edge computing.

[0010] Compared with the prior art, the present invention provides a method and system for managing low-altitude economic flight data based on edge computing, which has the following beneficial effects: This solution deploys edge computing nodes on the aircraft and ground base stations to collect multi-source data such as device status, communication signals, and environmental parameters in real time. Through denoising, normalization preprocessing, and a circular queue caching mechanism, efficient data management is achieved, and the end-to-end delay is reduced to within 50 ms, solving the real-time bottleneck caused by cloud transmission. This solution combines the composite biometric code that binds the device hardware fingerprint and the user operation habit, stores the data blocks in multiple edge nodes in a fragmented manner through the dynamic chaotic encryption algorithm, and introduces a dual verification mechanism (feature code verification + secondary authentication) to build a data security barrier against leakage and tampering, meeting the requirements of privacy protection regulations. This solution constructs a flight trajectory and energy consumption simulation model based on digital twin technology, dynamically adjusts the allocation of resources such as computing power and bandwidth through an adaptive entropy weight algorithm, and combines a task priority queue (such as giving priority to emergency tasks), resulting in a 30% improvement in resource allocation efficiency and a computing power utilization rate of over 90%. At the same time, an LSTM neural network is used to predict power consumption parameters, and the main power supply and backup power supply are coordinated to implement a stepped power supply strategy, extending the flight duration of the aircraft by 15%-20%. This solution models through the health status coefficient, conducts abnormal fluctuation clustering analysis and historical baseline comparison, warns of potential equipment failures 2 hours in advance, and reduces maintenance costs by 40%. Moreover, a user operation baseline is established based on unsupervised learning to block abnormal access in real time and update the defense strategy library, realizing the self-evolution of security strategies.

[0011] This solution forms a closed-loop management system of "data collection - secure transmission - intelligent analysis - dynamic optimization", breaks through the performance and security bottlenecks of traditional cloud computing, provides core technical support with low latency, high reliability, strong security, and low energy consumption for low-altitude scenarios such as drone logistics and emergency rescue, and significantly improves the real-time decision-making ability and system robustness in the low-altitude economic scenario. Brief Description of the Drawings

[0012] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0013] Figure 1 It is a block diagram of the system hierarchical architecture of the present invention; Figure 2 It is a waterfall flowchart of the method steps of the present invention; Figure 3 It is a flowchart of data encryption and storage of the present invention; Figure 4 It is a flowchart of intelligent power management of the present invention; Figure 5 It is a state diagram of abnormal behavior monitoring of the present invention. Detailed Embodiments

[0014] To make the purpose, technical solutions, and advantages of the present application clearer, the following will further describe the embodiments of the present application in detail in conjunction with the drawings.

[0015] Aiming at the problems existing in the management of flight data in the low-altitude economy, such as high real-time processing latency, prominent data security risks, inefficient resource scheduling, lagging fault prediction, and extensive energy consumption management.

[0016] This solution proposes a method for managing low-altitude economic flight data based on edge computing.

[0017] By deploying edge computing nodes, multi-source data such as device status and environmental parameters are collected in real time to achieve efficient data management, reducing the end-to-end latency to within 50 ms and solving the real-time bottleneck caused by cloud transmission. Using a composite biometric code that binds the device hardware fingerprint to the user's operation habits, a data security barrier against leakage and tampering is constructed to meet the requirements of privacy protection regulations. Based on the flight trajectory and energy consumption simulation model, the computing power is dynamically adjusted through the adaptive entropy weight algorithm, increasing the resource allocation efficiency by 30% and the computing power utilization rate to over 90%. Through health status coefficient modeling, abnormal fluctuation clustering analysis, and comparison with historical baselines, potential device failures are warned 2 hours in advance, reducing the maintenance cost by 40% and achieving self-evolution of security policies.

[0018] This solution forms a closed-loop management system of "data collection - secure transmission - intelligent analysis - dynamic optimization", breaking through the performance and security bottlenecks of traditional cloud computing, providing core technical support with low latency, high reliability, strong security, and low energy consumption for low-altitude scenarios such as drone logistics and emergency rescue, and significantly improving the real-time decision-making ability and system robustness in low-altitude economic scenarios.

[0019] Example 1, as Figures 1 - 5 shown, an illustrative description of the method for managing low-altitude economic flight data based on edge computing provided in the embodiments of this application includes the following steps: Step S100: Through the edge nodes of the aircraft and the ground base station, device status data such as IMU and BMS, and environmental parameters such as 2.4 GHz / 5.8 GHz communication signals and Beidou positioning are collected at a frequency of 100 Hz. After denoising by Kalman filtering and removing outliers by the IQR algorithm, Z-score normalization processing is performed. Using a circular data queue, it is segmented and cached in 500 ms windows, and redundant data is dynamically cleared based on the LRU algorithm. When the cache occupancy rate exceeds 80%, non-critical segments 24 hours ago are preferentially deleted. Step S200: A 128-bit composite biometric code is generated by fusing the device MAC address, control pressure curve, etc. Data blocks are dynamically encrypted through the Logistic chaos algorithm and stored in a distributed manner via IPFS according to the network latency (1 KB sharding + 1 copy for ≤50 ms, variable sharding + 3 redundant shards for >100 ms). When accessing, the feature code is first verified. If the Hamming distance > 5 bits, a SMS verification code or iris recognition secondary verification is triggered. Step S300: Calculate the health coefficient H (H = 0.5S_d / 10 + 0.3S_c + 0.2S_e) based on the device status (engine / battery / attitude score), communication quality, and environmental risk, and analyze the parameter change rate in combination with a sliding window. Cluster abnormal data through DBSCAN, and use the Weibull model to predict the fault type and remaining time. A yellow warning is triggered when H < 0.6, and a red alarm is activated when H < 0.4; Step S400: Build a lightweight twin model in the Unity engine, simulate the trajectory and energy consumption at a step size of 20 ms, and trigger the adaptive entropy weight algorithm when the deviation exceeds 5%. Calculate the weights of computing power / bandwidth / electricity according to the task urgency (emergency rescue weight 1.0), and schedule resources across nodes through the Hungarian algorithm. An additional 20% of computing power is dynamically allocated for high-priority tasks, and the scheduling delay < 100 ms; Step S500: Input the current RMS, voltage ripple, etc. into a 3-layer LSTM network to predict the power consumption (error ≤ 5%). The virtual power manager supplies power according to the load level: for low load (≤ 50W), only use the supercapacitor; for medium load (50 - 150W), the main power supply is connected in parallel with the capacitor; for high load (> 150W), the capacitor provides pulsed current. When the temperature of the lithium battery > 45°C, the current is reduced to 0.5C, and the battery life is extended by more than 15%; Step S600: Establish an operation baseline through the isolation forest algorithm, and monitor anomalies such as sensitive data access and high-frequency operations during non-working hours. When an anomaly is detected, level 1 records the log, level 2 blocks the connection, and level 3 starts isolation and updates the defense policy library. The anomaly response time < 500 ms, and the detection rate of new attacks increases by 3 - 5% per week to ensure the credibility and controllability of data operations.

[0020] As Figure 2 shown, the step S100 is used for edge node data acquisition and cache management, providing raw materials for subsequent analysis, including: S110: S1101: Construction of the device status data link: Connect the engine ECU, BMS battery management system, and IMU inertial measurement unit through the CAN bus (baud rate 1 Mbps), adopt the master-slave synchronization mechanism (Master is the flight control computer, Slave is the sensor node), and send a synchronization pulse (SyncPulse) every 10 ms to calibrate the sampling clock; Define the binary data frame format, including: Frame header (0xAA55, 2B), timestamp (64-bit UTC nanosecond value, 8B), sensor ID (1B, 0x01 - 0x05 correspond to IMU / ECU / BMS, etc. respectively), data body (variable length, e.g., IMU contains 3-axis acceleration / angular velocity, a total of 18B), CRC checksum (2B). The IMU / barometer operates at 100Hz, and the BMS / ECU operates at 20Hz, meeting the real-time requirements of flight control; S1102: RF front-end design: Adopt a software-defined radio architecture based on RTL-SDR (software-defined radio), covering the 2.4GHz / 5.8GHz UAV communication frequency band, with a sampling rate of 2MSPS, and perform real-time calculations through FPGA (Xilinx Zynq series); Received Signal Strength Indicator (RSSI, unit: dBm): RSSI = 10 × (Power spectral density × channel bandwidth); Round-trip time delay (RTT, unit: ms): By sending ICMP Echo Request and recording the response time, take the average of 5 samples; Signal-to-Noise Ratio (SNR, unit: dB): SNR = 10 × , and the noise power is estimated by sampling the idle channel; S1103: Adopt a high-precision barometer (accuracy ±0.1hPa, altitude conversion error ±1m), temperature and humidity sensor (accuracy ±0.5℃ / ±2%RH), ultrasonic anemometer (range 0 - 30m / s, accuracy ±0.1m / s), and access the edge node through the I2C bus (rate 400kHz); At the same time, receive GPS L1 (1575.42MHz) and Beidou B1 (1561.098MHz) signals, fuse multi-source positioning data through Extended Kalman Filter (EKF), and output longitude (accuracy ±1m), latitude (accuracy ±1m), and altitude (accuracy ±2m); For example: In the UAV logistics scenario, IMU data is used for real-time attitude solution to ensure the smooth transportation of goods; RSSI and SNR data are used to dynamically switch communication channels (such as switching from 2.4GHz to 5.8GHz to avoid interference) to ensure the reliability of command transmission.

[0021] S120: S1201: Time synchronization and correction: The edge node is equipped with a temperature-compensated crystal oscillator (TCXO, accuracy ±1ppm), which is calibrated by the NTP server once an hour; perform linear interpolation on the timestamps of the collected data to correct the clock deviation between the sensor and the edge node: Formula: = +Δt× , where Δt is the total drift during the calibration period; S1202: Outlier detection and repair: For each sensor channel (such as the X-axis acceleration of the IMU), calculate the quartiles Q1 and Q3, and determine IQR = Q3 - Q1; Define the outlier range: [Q1 - 1.5×IQR, Q3 + 1.5×IQR], and values outside this range are marked as outliers; Outlier repair: Use linear interpolation of the previous and next values (e.g., if the i-th data point is an outlier, then ); Convert the air pressure value from hPa to Pa (1 hPa = 100 Pa), the temperature from °C to K (K = °C + 273.15), and the acceleration from m / s² to g (1 g = 9.80665 m / s²); For each feature dimension (such as battery voltage, RSSI), calculate the mean μ and the standard deviation σ, and the normalized value is: , making the data distribution follow N(0, 1); Formula derivation: Assume a battery voltage sequence is [3.7, 3.8, 3.5, 4.0], calculate μ = 3.75, σ ≈ 0.176, then the normalized values are [0.28, 0.85, -1.42, 1.42], which is convenient for input to the neural network.

[0022] S130: S1301: Segment header (20B): Timestamp range (start / end nanosecond values, 8B each), data length (4B), reference count (1B), status bit (1B, 0 = idle, 1 = occupied, 2 = to be cleared); Data body (variable length, maximum 4KB): Store the preprocessed binary data; Initially create 10 empty segments, with a size of 1KB / segment; When a write request arrives, preferentially allocate a segment with the status of "idle". If there is no available segment, dynamically create a new segment (expand by 5 segments each time, and the size increases by 1.5 times that of the previous segment, with a maximum of 16KB / segment); S1302: Maintain a doubly linked list, with the head being the most recently used segment and the tail being the least recently used segment; When the cache occupancy rate > 80%, check the segment status starting from the tail: If the status is "to be cleared" and the reference count = 0, immediately delete and reclaim the memory; If the status is "occupied", mark it as "to be cleared", decrement the reference count by 1, and postpone the cleanup; Occupancy rate = ×100%, the total cache capacity is initially 10KB and can be expanded up to 128KB at most; S1303: Receive a data write request → Search for or create a data segment → Write the data body → Update the segment header timestamp and length → Move the segment to the head of the LRU linked list; Search for segments based on the timestamp range → Verify the timestamp validity → Copy the data body to the read buffer → Move the segment to the head of the LRU linked list → Return the data pointer.

[0023] Based on the preprocessed highly reliable data, the system needs to further build a security protection system to ensure the confidentiality and integrity of the data during storage and transmission. The step S200 is used for secure storage and trusted access after data collection and cleaning to protect the security of data assets. Specifically: S210: S2101: Hardware feature collection: Motherboard: Read the UUID, manufacturer, and model through the SMBIOS protocol; Sensors: Read the calibration coefficients of the IMU and the factory serial number of the barometer; Communication module: Obtain the MAC address of the Wi-Fi chip and the BD_ADDR of the Bluetooth module; Concatenate the above information into a string such as "UUID_ABC123_MAC_00:11:22:33:44:55", and generate a 64-byte device fingerprint (such as 0x1a2b3c...d4e5f6) through the SHA-512 algorithm; S2102: Operational data collection: Sample the voltage of the pressure sensor through the ADC (precision 12 bits, resolution 0.00024V), and record the curve of the pressure degree changing with time (sampling rate 20Hz); Monitor the instruction sending time interval of the ground station software (such as the time difference between two "flight path update" instructions), and the click order of the interface elements (such as first selecting "take a photo" and then selecting "waypoint 1"); Extract 12 statistical features such as mean, standard deviation, peak value, and rising edge slope; Use the DTW algorithm to calculate the similarity between the current operation sequence and the historical template, and generate a 10-dimensional feature vector; Concatenate the device fingerprint and the behavior feature vector, and generate a 128-byte composite biometric code through the HMAC-SHA256 algorithm (the key is the hash value of the user ID); Formula derivation: Let the device fingerprint be F, the behavior feature vector be B, and the user ID be U. Then the feature code C = HMAC-SHA256(K, F||B), where K = SHA256(U).

[0024] S220: S2201: Chaotic encryption algorithm integration: Standard Logistic map: +1 = μ (1 − ), it enters the chaotic state when μ ∈ (3.5699456, 4]; Dynamic parameter adjustment: Set μ = 3.8 + 0.1 × min( , 2) according to the network delay T (unit: ms) to ensure the improvement of encryption strength at high latency; S2202: Data sharding and distributed storage: Low-latency scenario (T ≤ 50 ms): Fix the shard size at 1 KB, and generate 1 redundant copy for each data block (total number of shards = original number of blocks × 2); High-latency scenario (T > 100 ms): Variable shard size (512 B - 4 KB), and use Reed-Solomon coding to generate 3 redundant shards (total number of shards = original number of blocks + 3); Generate a pseudo-random number r through VDF (Verifiable Random Function), and calculate the storage node index: node_id = (r mod N) + 1 (N is the number of available nodes); Call VDF every 10 minutes to regenerate r and update the storage nodes of all data shards to ensure dynamic anti-attack; When using Reed-Solomon(5, 3) coding, at most 3 node failures are allowed, and the data recovery probability is 100%. The calculation formula is: = 1 − ; ( is the single-node failure rate. Assume = 0.01, then ≈ 99.9999%) In the formula, k is the adaptive weight coefficient (value range 0.8 - 1.2), which is used to adjust the resource allocation priority.

[0025] S230: S2301: Feature code comparison: Extract the hardware fingerprint F′ and operation behavior characteristics B′ of the current device to generate the feature code C′; Calculate the Hamming distance: d = Hamming(C, C′). If d ≤ 5 (i.e., 5-bit binary bit differences are allowed), the verification passes; Query the permission table according to the user ID (e.g., "ordinary user" can only read flight data, "administrator" can modify task parameters), and return the allowed operation set; S2302: Dynamic verification code generation: SMS verification code: Generate a 6-digit random number (range 100000 - 999999), and send it to the registered mobile phone number through an encrypted channel (such as HTTPS), with a validity period of 30 seconds; Biometric recognition: Adopt minutiae feature matching of fingerprint images (such as extracting the coordinates of ridge endpoints and bifurcation points), calculate the matching score, and set the threshold to 70 / 100; Verification logic: When the primary verification fails, the secondary verification is automatically triggered. The user needs to complete the verification within 5 minutes, otherwise the account will be locked for 15 minutes; For example: In the emergency rescue scenario, when temporarily authorizing device access, if the primary verification fails due to hardware fingerprint differences, iris recognition secondary verification is triggered. After the rescue personnel pass, the system automatically grants the temporary permission of "real-time video stream reading", and the permission is immediately revoked after the task ends.

[0026] After completing the secure data storage, it is necessary to perform in-depth analysis on the data to evaluate the device operation status, identify potential risks in advance and trigger the warning mechanism, and convert the securely stored data into quantitative indicators of the device health status to support preventive maintenance.

[0027] Specifically, the step S300 is used for real-time status evaluation and fault prediction, and mining data value to identify risks, including: S310: S3101: Device status scoring system: The value range is 0 - 10 points, calculated according to the rotational speed volatility (rpm_var = ) = ; Calculate the remaining capacity (SOC) through the ampere-hour integration method, combined with the internal resistance growth rate (R_delta = ) = 8× + 2×(1 - min(R_delta, 0.2)); Calculate the standard deviation of the IMU attitude angle ( ), scoring formula: = 10× (unit: degree); S3102: Communication quality scoring: = 0.6× + 0.4×(1 - ); Among them, = -100dBm, , = 200ms; = 0.5 × min( ) + 0.5 × (1 - ). When the distance to the obstacle is > 50 meters, the score for this part is 0.5; S3103: Health factor synthesis: H = 0.5 + 0.3 + 0.2 (where = ). The weights are determined by logistic regression training of historical failure data, and the AUC value reaches 0.92.

[0028] S320: S3201: For the battery voltage sequence V = , ,..., , when the window size N = 50, calculate the average descent rate; slope = (Δt is the sampling interval, 0.05 s), (Δt is the sampling interval, 0.05 s); Calculate the mean value of the voltage descent rate in the past 7 days and the standard deviation . If the current slope < − , mark it as "battery degradation trend"; S3202: Calculate the Z value for the standardized data, and mark the points with |Z| > 3 as potential anomalies; Define the neighborhood radius ϵ = 0.5 (Euclidean distance), and the minimum number of samples MinPts = 3; Cluster the potential anomaly points into several classes, such as "communication interruption class" (including RSSI < -90 dBm, RTT > 500 ms), "power anomaly class" (engine speed volatility > 0.2); For each cluster, calculate the proportion of the number of samples within the class , and the influencing factor is × the distance from the cluster center to the origin, which is used to sort the anomaly priorities; For example: When a certain drone is flying at the 30th minute, the battery voltage descent rate slope = -0.02 V / s, which is lower than the baseline mean of -0.01 V / s and exceeds 2σ. At the same time, DBSCAN detects that 5 consecutive points belong to the "power anomaly class", triggering a "battery thermal runaway risk" warning. The actual failure occurs 1 hour and 45 minutes later, and the warning lead time reaches 2 hours and 15 minutes.

[0029] S330: S3301: Fault type classifier: Feature input: health coefficient H, trend analysis coefficient (such as slope normalization value), abnormal influence factor set; Model architecture: 3-layer fully connected neural network, 10-dimensional input layer, 20-dimensional and 10-dimensional hidden layers respectively, output layer corresponding to fault types (such as battery fault, communication fault, sensor fault), using Softmax activation, and the loss function is cross-entropy; Collect more than 5000 fault samples, the positive-negative sample ratio is 1:3, use the Adam optimizer, learning rate 0.001, and train until the accuracy of the validation set reaches 85%; S3302: Adopt the Weibull proportional hazard model, and the formula is: h(t)= (t)exp( + +...+ ); Among them, h(t) is the failure rate, is the feature (such as H, slope), is the regression coefficient, which is determined by maximum likelihood estimation; When the prediction is a battery fault, calculate the survival function S(t)=exp(− , and take the t value when S(t)=0.1 as the remaining available time; S3303: Alarm generation and distribution: The alarm levels are divided into three levels (yellow / orange / red), and the corresponding H value ranges are [0.4,0.6), [0.2,0.4), <0.2 respectively; The red alarm is simultaneously pushed to the flight control system (triggering automatic return), ground station operator (SMS + voice), and operation and maintenance center (email + system pop-up window) to ensure that it is reached within 5 seconds.

[0030] Based on the real-time status evaluation results, the system needs to dynamically optimize the resource allocation strategy to adapt to the rapid changes of flight missions and environments. The status evaluation and fault prediction provide a data basis for resource optimization. It is necessary to introduce simulation and intelligent algorithms to achieve dynamic scheduling to meet dynamic task requirements.

[0031] The step S400 is used for digital twin-driven resource optimization to improve the system response efficiency, including: S410: S4101: Multi-physical field modeling: Dynamic model: ; Among them, m is the mass of the aircraft, is the thrust, D(v)=0.5 , D(v) is the air resistance, (ρ is the air density, is the drag coefficient, and A is the frontal area); Energy consumption model: = I × V × Δt + × d (current I, voltage V, flight distance d); S4102: Extract the first 10 modes using Proper Orthogonal Decomposition (POD), reduce the state space from 6 dimensions (position x / y / z, velocity v_x / v_y / v_z) to 3 dimensions, and reduce the computational amount by 70%; Render multiple drone twins using GPU instancing in the Unity engine, reduce the number of batch DrawCalls by 90%, and stabilize the frame rate at 60 FPS; S4103: The edge node sends a status update (including position, velocity, and attitude) every 20 ms, and the digital twin model synchronizes the intermediate frames through linear interpolation to ensure that the visual delay < 50 ms.

[0032] S420: S4201: Normalize the computing power demand C, bandwidth demand B, and remaining power E to the interval [0, 1]; = , = , = ; For each metric, calculate the probability of the i-th task = , entropy value: = − ln ; Weight = , where m = 3 (computing power / bandwidth / electricity); S4202: = Urgency × × + Importance × × + Timeliness × × ; Urgency value: Emergency rescue (10) > Medical transportation (8) > General logistics (5); Adopt preemptive priority scheduling, and high-priority tasks can interrupt the resource allocation of low-priority tasks to ensure that the delay of critical tasks < 50 ms; S4203: After each resource allocation, calculate the deviation between the actual effect and the simulation result (such as the deviation of computing power utilization rate ΔC = ∣ − | / ) If ΔC > 0.1, then = + 0.1×ΔC, and converges to a stable value through 5 iterations; Specifically, the entropy weight method determines the weight through the variability of indicators, avoiding the deviation of artificial assignment. When the computing power demand fluctuates greatly, its entropy value decreases and the weight automatically increases. For example, the standard deviation of the computing power demand of an emergency task is 3 times that of an ordinary task, and its computing power weight increases from 0.3 to 0.5, meeting the actual demand.

[0033] S430: S4301: Real-time collect the CPU utilization rate (%), GPU memory occupancy (MB), and remaining communication bandwidth (Mbps) of each node, and report them to the central manager through the gRPC protocol; Construct the global resource matrix R = , where n is the number of nodes, m is the resource type (CPU / GPU / bandwidth), is the remaining amount of resource j of node i; S4302: Convert the task allocation into a bipartite graph matching problem. The element of the cost matrix C is the processing delay of task k on node i (formula: = + is the task data volume, is the CPU frequency of the node, is the node bandwidth); When the CPU utilization rate of a certain node > 85%, trigger load migration, and migrate the unfinished part of the low-priority task to the node with a utilization rate < 30%. The migration overhead is calculated through the estimated data volume and bandwidth ( = + , S is the size of the status data); S4303: Allocate an exclusive token bucket for high-priority tasks (such as the token generation rate of the emergency task is 100Mbps, and the bucket capacity is 500MB) to ensure that the bandwidth is preferentially allocated during burst traffic; Define three levels of quality of service: real-time control (delay < 10ms), video stream (delay < 100ms), and log transmission (best effort), and implement differential services through DSCP marking.

[0034] One of the goals of resource optimization is to improve the energy usage efficiency. Therefore, it is necessary to further combine power consumption prediction to achieve intelligent management of the power system. The optimization of resource allocation needs to be coordinated with energy management to balance performance requirements and battery life.

[0035] The step S500 is used for intelligent power management and energy consumption optimization to extend the battery life of the device, including: S510: S5101: Build a high-precision power consumption prediction model: The root mean square (RMS) value of the current, the voltage ripple factor (γ = ), and the one-hot encoding of the task type (a 3D vector, e.g., logistics = 100, inspection = 010, rescue = 001); Output labels: average power (W) and peak power (W) in the next 15 minutes; Use Min-Max normalization to scale the current / voltage features to [0, 1]. The formula is: = ; S5102: A 3-layer LSTM (64 neurons per layer, dropout rate 0.2) + 2-layer fully connected (32D and 1D), with activation functions tanh and linear respectively; Weighted sum of mean absolute error (MAE) + root mean square error (RMSE). The formula is: Loss = 0.7×MAE + 0.3×RMSE; Batch size 32, number of epochs 50, learning rate 0.0001. Use early stopping (patience = 10) to prevent overfitting. The MAE of the validation set is 4.2W and the RMSE is 5.8W; S5103: After each flight mission, add 500 newly collected samples to the training set and retrain the model to ensure that the model adapts to power consumption drift caused by equipment aging or environmental changes.

[0036] S520: S5201: Power status monitoring: Monitoring parameters include SOC (accuracy ±2%), internal resistance (accuracy ±5mΩ), temperature (accuracy ±1°C). When SOC < 20%, it is marked as "low battery status"; Monitor the capacitor voltage (accuracy ±0.1V) and equivalent series resistance (ESR, accuracy ±10mΩ). When the voltage < 80% of the rated value, start charging; S5202: Step-by-step power supply strategy: Low-load mode: Only the supercapacitor supplies power, with a current limit ≤ 3A to extend the life of the lithium battery; When the supercapacitor voltage < 2.5V (rated 3V) and the main power supply SOC > 80%, charge with a current of 1A; Medium-load mode: The main power supply and the supercapacitor supply power in parallel. The current distribution formula is: Ibattery = I× ( is the internal resistance of the main power supply, is the ESR of the supercapacitor); High-load mode: The main power supply outputs at full capacity, and the supercapacitor provides instantaneous pulsed current (up to 10 A) for a duration of ≤ 5 seconds to prevent the lithium battery from being damaged due to overcurrent; S5203: When the temperature of the main power supply > 45 °C, automatically reduce the charging current to 0.5C and switch to the medium-load mode, preferentially using the supercapacitor until the temperature drops below 40 °C; Energy efficiency calculation: Assume the energy density of the lithium battery is 200 Wh / kg and that of the supercapacitor is 10 Wh / kg. In the low-load mode, the hourly energy consumption is 50 Wh, of which the supercapacitor undertakes 40 Wh (accounting for 80%), and the lithium battery only consumes 10 Wh, saving 80% of energy compared to the all-lithium battery solution. The equivalent extended endurance time: Δt = × Endurance improvement factor (m is the battery mass, and the factor is taken as 1.2); S530: S5301: Combine the power consumption prediction model and use the A* algorithm to calculate the minimum energy consumption path. The constraint conditions are that the maximum flight speed ≤ 15 m / s and the obstacle avoidance distance ≥ 2 meters; Place heavy objects as close as possible to the center of gravity of the aircraft to reduce the energy consumption for attitude adjustment. The formula: ΔE = 0.5 × J × ( − )(moment of inertia J, angular velocity change ω − ); S5302: According to the real-time power consumption prediction, adopt an economic throttle (the throttle corresponding to the highest motor efficiency is 50% - 60%) during the level flight phase and temporarily increase it to 70% during the climbing phase; Enable motor reverse braking during the landing phase to convert kinetic energy into electrical energy and store it in the supercapacitor. The recovery efficiency is about 25% - 30%; S5303: Adopt a three-stage charging method of "constant current - constant voltage - trickle charge". When the SOC > 90%, the current drops to 0.1C to prevent overcharging and battery aging; When the aircraft is stationary for more than 24 hours, it automatically enters the sleep state, the power consumption drops to < 1 mW, and the wake-up time < 500 ms.

[0037] After completing data processing, resource optimization, and energy management, it is necessary to construct a safety monitoring closed-loop to prevent human misoperation or malicious attacks. The stability and security of the system depend on the real-time monitoring and defense of operation behaviors.

[0038] The step S600 is used for abnormal behavior monitoring and security reinforcement to consolidate the system reliability, including: S610: S6101: User ID, operation time (accurate to milliseconds), data type (such as flight route data / task instructions / sensor parameters), operation type (read / modify / delete), file path depth (such as the depth of " / data / flight_logs / 2025 / 05" is 3); Use UUID to mark user sessions and record context information such as session duration, operation frequency, data transfer volume, etc.; S6102: Randomly select features (such as operation frequency, path depth), randomly generate a separation hyperplane in the feature space, and calculate the height of the "isolation tree" for each sample; Anomaly score formula: s(x)= , where E(h(x)) is the average tree height and c(n) is the expected tree height of n samples; Set a threshold =0.9, and a score > 0.9 is considered an anomaly; S6103: Retrain the model on the operation logs of the previous 24 hours every day at midnight, update the normal behavior boundary, and adapt to the operation mode differences between weekdays / weekends (such as a 30% decrease in operation frequency on weekends); S620: S6201: Users download sensitive data (such as files under the path " / system / waypoints / ") during non-working hours (22:00 - 6:00), and the transfer volume > 10MB; Formula: alert = (time period ∈ non-working) ∧ (path depth ≥ 4) ∧ (transfer volume > 10MB); Monitor ordinary users' attempts to perform administrator operations (such as modifying flight control parameters), match operation permissions through an access control list (ACL), and the probability of misoperation < 0.1%; S6202: Through the ARP spoofing detection algorithm, compare the IP-MAC binding table with real-time ARP packets. If the same IP corresponds to different MAC addresses and occurs in a non-switching scenario, it is determined as a spoofing attack; Use a machine learning model to train the metadata of encrypted traffic (such as packet size distribution, connection duration) to identify hidden C2 communications (Command and Control), with an accuracy of 88%; S6203: Hierarchical response: Level 1 anomaly (such as misoperation): Record the log and send a warning to the user; Level 2 anomaly (such as unauthorized access): Block the current connection and generate an event work order to push to the security team; Level 3 anomaly (such as ransomware attack): Trigger a full-system isolation, shut down non-essential services, and start the data recovery process; S630: S6301: Parse the features of the attack samples from the defense strategy library, such as IP address segments (e.g., 192.168.1.100 - 192.168.1.200), operation modes (e.g., 10 consecutive failed login attempts), and protocol fields (e.g., the HTTP request header contains ".. / .. / "); Use the TF-IDF algorithm to vectorize the features and construct an attack feature vector space; S6302: Whenever a new attack type is detected, add its feature vector to the training set and use One-Class SVM to dynamically expand the anomaly detection boundary; Simulate attack scenarios (e.g., brute force cracking, SQL injection), generate adversarial samples and inject them into the model to improve robustness, and the success rate of adversarial samples drops from 30% to 15%; S6303: Send the updated detection rules to all edge nodes through a security policy management platform (e.g., Fortinet Security Fabric), and use hash verification to ensure rule consistency; Perform a simulated attack drill once a week to verify the effectiveness of the defense strategy, record the drill results and generate an improvement report; So far, the system forms a complete closed loop of 'data collection - secure storage - status assessment - resource optimization - energy consumption management - security monitoring', realizing the intelligent management of the entire life cycle of low-altitude flight data.

[0039] Experimental example: Experimental purpose Verify the real-time performance, security, and energy efficiency of the low-altitude flight data management method based on edge computing in unmanned aerial vehicle logistics.

[0040] Hardware: DJI Matrice 300 RTK unmanned aerial vehicle (equipped with edge computing node NVIDIA Jetson Xavier NX), ground base station (Intel i7 - 12700H, 16GB RAM); Sensors: IMU (Bosch BMI085), barometer (MS5611), radio frequency sensor (RTL-SDR); Software: ROS2 Humble, Unity2021.3, Python 3.9 (integrated with PyTorch / LSTM / Scikit-learn).

[0041] Step 1: The unmanned aerial vehicle carries 5 kg of goods and executes a 3-kilometer logistics route. The edge node collects IMU data (such as acceleration , , ), BMS battery voltage ( ) and RSSI signal strength at 100 Hz; Apply Kalman filter denoising to the original data and normalize it to the [-1, 1] interval through Z-score; Use a circular queue to cache data, set the cache threshold to 80%, and automatically clean historical data older than 24 hours; Experimental data record form of sensor original data and preprocessing results: Step 2: Extract the pressure curve features of the drone MAC address (00:12:34:AB:CD:EF) and the pilot's control handle, and generate a 128-bit feature code through SHA-256; Encrypt the data block using Logistic chaotic encryption (μ = 3.8) and store it in 3 edge nodes through IPFS sharding; Simulate unauthorized device access, trigger secondary verification (SMS verification code), and record the verification time consumption.

[0042] Step 3: Calculate the health status coefficient H: device status score = 8.2, communication quality score = 0.75, environmental risk score = 0.6, comprehensive H = 0.5×8.2 / 10 + 0.3×0.75 + 0.2×0.6 = 0.745; Detect that the battery voltage drop rate for 3 consecutive windows is slope = -0.015 V / s (lower than the baseline of -0.01 V / s), and trigger the "battery degradation trend" warning; Experimental data record form of health status score and warning parameters: Step 4: Build a digital twin model in Unity, simulate and predict a flight trajectory deviation of 4.2 meters (threshold 5 meters), and trigger the adaptive entropy weight algorithm; Calculate the weights of computing power / bandwidth / electricity as 0.5 / 0.3 / 0.2, allocate an additional 15% computing power for the logistics task, and the adjusted trajectory deviation is reduced to 0.8 meters.

[0043] Step 5: The LSTM model predicts the average power in the next 15 minutes to be 118 W (actual value 122 W, error 3.3%); Adopt "main power + super capacitor" parallel power supply. When the load is low, the super capacitor undertakes 70% of the current, and the endurance is extended from 45 minutes to 55 minutes; Experimental data record form of resource allocation weights and power consumption comparison: Step 6: Simulate a user downloading sensitive flight route data during off-peak hours (23:00), with a transmission volume of 12 MB, triggering a first-level anomaly alert; Block the connection and record the characteristics in the defense strategy library. The anomaly detection takes 470 ms; Experimental data record form for security verification and response time: Example 2: An edge-computing-based low-altitude economy flight data management system for implementing an edge-computing-based low-altitude economy flight data management method, including the following modules; Edge data collection and preprocessing module: Through edge computing nodes deployed on aircraft and ground base stations, it collects multi-source data such as device status, communication signals, and environmental parameters in real time, performs denoising and standardization preprocessing, and adopts a circular data queue caching mechanism to manage data in segments according to time windows, and dynamically clears redundant data according to cache occupancy; Distributed secure storage and verification module: Generates a composite biometric code containing device hardware fingerprints and user operation characteristics, encrypts data blocks in slices using a dynamic chaotic encryption algorithm, and stores them in multiple edge nodes through a distributed hash table; When accessing, first verify the feature code, and trigger SMS verification code or biometric secondary verification in case of anomalies; Intelligent status evaluation and prediction module: Calculates device health status coefficients and trend analysis coefficients based on multi-dimensional data, detects abnormal fluctuations through Z-score and DBSCAN algorithms, generates sub-fluctuation impact factors, triggers a fault warning when exceeding the threshold, and outputs the fault type and remaining time; Digital twin-driven resource optimization module: Builds a lightweight digital twin model to simulate flight trajectories and energy consumption. If the deviation exceeds 5%, it triggers an adaptive entropy weight algorithm, dynamically allocates computing power and bandwidth resources according to the task urgency, and generates a priority queue; Intelligent power management module: Uses an LSTM neural network to predict power consumption parameters, coordinates the main power supply and supercapacitors through a virtual power manager, executes a stepped power supply strategy, and optimizes power distribution in combination with a Bayesian network to extend the battery life; Abnormal behavior monitoring module: Establishes a user operation baseline through unsupervised learning, real-time monitors indicators such as access frequency and path depth, blocks the connection when an anomaly is detected, records the characteristics in the defense strategy library, and updates the security model.

[0044] Example 3: A computer device, the computer device includes: a processor and a memory, and at least one instruction, at least one program, a code set or an instruction set is stored in the memory, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement an edge-computing-based low-altitude economy flight data management method.

[0045] Embodiment 4: A computer-readable storage medium, characterized in that at least one instruction, at least one program, a code set or an instruction set is stored in the storage medium, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by a processor to implement a method for managing low-altitude economic flight data based on edge computing.

[0046] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for managing low-altitude economic flight data based on edge computing, characterized in that Including: S100: Collect multi-source data of low-altitude flight devices, including power consumption characteristic data, through edge computing nodes deployed on aircraft and ground base stations, perform preprocessing operations on the data in the edge computing nodes, and at the same time use a circular data queue caching mechanism to manage data in segments, configure dynamically allocated cache space for each data segment, and perform data storage or cleaning operations according to the occupancy of the cache space; S200: Generate a composite biometric code for the flight data in the cache, use a chaotic encryption algorithm to fragment and encrypt the data blocks, distribute and store the encrypted data blocks in multiple edge nodes, verify whether the composite biometric code is consistent with the stored feature code on the edge computing node, if not, start a secondary verification mechanism to further confirm the user identity; S300: Extract the abnormal fluctuation characteristics of the flight data, calculate the device health status coefficient and trend analysis coefficient based on the preprocessed data. At the same time, combine historical flight data and multi-dimensional parameters collected in real time to predict the future state of the aircraft, and trigger a fault warning if it exceeds the threshold; S400: Build a digital twin model based on the current flight data and resource allocation scheme, and predict the flight trajectory and energy consumption through a physical simulation engine; if the deviation between the simulation result and the preset threshold exceeds 5%, trigger an adaptive entropy weight hybrid optimization algorithm, and generate a priority queue according to the flight task urgency and resource weights to dynamically adjust the resource allocation strategy and iteratively adjust the resource allocation parameters; S500: Input the power consumption characteristic data of the aircraft into a heterogeneous device management neural network for power management prediction, generate device power consumption prediction parameters, and perform power response control on the main power supply and standby power supply of the aircraft through a virtual power manager to generate power adjustment parameters; S600: Monitor the user's operation behavior on the aircraft data. When an abnormal behavior is detected, record the behavior and store it in the defense policy library for subsequent security policy optimization; at the same time, use the real-time monitoring ability of the edge computing node to quickly respond to abnormal behaviors to ensure the security and reliability of the aircraft data.

2. The method for managing low-altitude economic flight data based on edge computing according to claim 1, wherein: The collection of the multi-source data includes: collecting the three-axis acceleration and angular velocity data of the IMU at a frequency of 100Hz through the CAN bus, real-time monitoring the RSSI signal strength and RTT transmission delay in the 2.4GHz / 5.8GHz frequency band through the RTL-SDR radio frequency sensor, and obtaining positioning data with an accuracy of ±1m through the Beidou module; the preprocessing operations include applying Kalman filtering to the IMU data for denoising, removing outliers through the IQR algorithm, and using Z-score normalization to map the data to the [-1,1] interval.

3. The method for managing low-altitude economic flight data based on edge computing according to claim 1, wherein: The composite biometric code is generated through the SHA-256 algorithm and contains hardware fingerprints such as the device MAC address and sensor serial number, as well as 12-dimensional statistical features of the control handle pressure curve; the chaotic encryption uses the Logistic map, and the parameter μ is dynamically adjusted according to the network delay. When the delay > 100ms, μ = 3.9, and the data fragmentation uses Reed-Solomon(5,3) coding to support 3-node fault tolerance.

4. The method for managing low-altitude economic flight data based on edge computing according to claim 1, wherein: The calculation formula of the health status coefficient H is: H = 0.5×(S_engine + S_battery + S_att) / 3 + 0.3×S_c + 0.2×S_e, where S_engine is the engine status score (0 - 10 points), S_c is the communication quality score (0 - 1), S_e is the environmental risk score (0 - 1), S_battery is the battery status score (0 - 10 points), S_att is the attitude sensor status score (0 - 10 points), and a yellow warning is triggered when H < 0.

6.

5. The method for managing low-altitude economic flight data based on edge computing according to claim 1, wherein: The digital twin model is built using the Unity engine, with a simulation step size of 20 ms, a trajectory prediction error ≤ 1 m, and an energy consumption prediction error ≤ 5%; the adaptive entropy weight algorithm calculates the weights of computing power, bandwidth, and power through the entropy value method, and the formula is ω_j = (1 - e_j) / Σ(1 - e_j), where e_j is the entropy value of the j-th index.

6. The method for managing low-altitude economic flight data based on edge computing according to claim 1, wherein: The neural network includes a 3-layer 64-neuron network layer, with input features being the current RMS, voltage ripple coefficient, and one-hot encoding of the task type, and the output power prediction error ≤ 5%; the power supply switching delay < 5 ms, and the supercapacitor bears ≥ 70% of the current under low load.

7. The method for managing low-altitude economic flight data based on edge computing according to claim 1, characterized in that: The isolation forest algorithm is used to establish a behavior baseline for monitoring the user's operation behavior on the aircraft data, with a neighborhood radius ε = 0.5 and a minimum number of samples MinPts = 3, and an alarm is triggered when the anomaly score s(x) > 0.9; The three-level response mechanism includes: the first-level abnormal record log, the second-level abnormal connection block, and the third-level abnormal start system isolation.

8. An edge-computing-based low-altitude economy flight data management system for implementing the method according to any one of claims 1-7, characterized in that, It includes the following modules: Edge data acquisition and preprocessing module: Through the edge computing nodes deployed on the aircraft and the ground base station, it collects multi-source data such as device status, communication signals, and environmental parameters in real time, performs denoising and standardization preprocessing, and adopts a circular data queue caching mechanism to manage data in segments according to time windows, and dynamically clears redundant data according to the cache occupancy; Distributed secure storage and verification module: Generates a composite biometric code containing device hardware fingerprints and user operation characteristics, encrypts data blocks in slices using the dynamic chaos encryption algorithm, and stores them in multiple edge nodes through a distributed hash table; when accessing, first verify the feature code, and trigger SMS verification code or biometric secondary verification in case of anomalies; Intelligent status evaluation and prediction module: Calculates the device health status coefficient and trend analysis coefficient based on multi-dimensional data, detects abnormal fluctuations through the Z-score and DBSCAN algorithms, generates sub-fluctuation impact factors, triggers a fault warning when exceeding the threshold, and outputs the fault type and remaining time; Digital twin-driven resource optimization module: Builds a lightweight digital twin model to simulate the flight trajectory and energy consumption. If the deviation exceeds 5%, it triggers the adaptive entropy weight algorithm, dynamically allocates computing power and bandwidth resources according to the task urgency, and generates a priority queue; Intelligent power management module: Uses the LSTM neural network to predict power consumption parameters, coordinates the main power supply and the supercapacitor through a virtual power manager, executes a stepped power supply strategy, and optimizes power distribution in combination with the Bayesian network to extend the battery life; Abnormal Behavior Monitoring Module: Establish a user operation baseline through unsupervised learning, monitor indicators such as access frequency and path depth in real time, block the connection when an anomaly is detected, record the features in the defense strategy library, and update the security model.

9. A computer device, characterized in that, The computer device includes: a processor and a memory. At least one instruction, at least one program, a code set, or an instruction set is stored in the memory. The at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the edge computing-based low-altitude economy flight data management method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, At least one instruction, at least one program, a code set, or an instruction set is stored in the storage medium. The at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by a processor to implement the edge computing-based low-altitude economy flight data management method according to any one of claims 1 to 7.

Images