Unmanned aerial vehicle nest full life cycle management and control method, system, product and equipment

Abstract

The invention discloses an unmanned aerial vehicle nest full life cycle management and control method, system, product and device, and relates to the field of unmanned aerial vehicle nests, and the method comprises the steps: employing a bidirectional access mechanism to carry out the bidirectional validity verification of an unmanned aerial vehicle and a nest, and generating an access safety certificate; integrating multi-dimensional data through edge computing nodes, and constructing a visual relation model and a dynamic behavior monitoring model; carrying out real-time risk identification and early warning by adopting a multi-dimensional feature fusion anomaly detection algorithm; the digital twinning technology is utilized to fuse aircraft nest operation data, and predictive maintenance is achieved through a life prediction model; and safely erasing the sensitive data of the equipment and cancelling the voucher to form a management closed loop. Through a bidirectional access mechanism, multi-dimensional anomaly detection, predictive maintenance and safe decommissioning processing, full-link active safety protection from access, operation, maintenance to decommissioning is realized, the safety and reliability of a machine nest system are improved, and the system can be applied to complex scenes with strict safety requirements such as electric power inspection.

Description

Technical Field

[0001] This invention relates to the field of drone nests, and in particular to a method, system, product, and equipment for the full lifecycle management of drone nests. Background Technology

[0002] Intelligent drone hangars (also known as drone base stations or hangars) serve as platforms for the automatic take-off and landing, charging, data exchange, and storage of drones, and their applications are becoming increasingly widespread. Existing drone hangar management solutions have evolved from basic mission support to full lifecycle management. For example, existing technologies (such as CN118710250A) have proposed a full lifecycle management system for drone hangar equipment based on digital twins. By constructing virtual models, they achieve fault diagnosis, performance prediction, and maintenance decisions, making some progress in improving the management efficiency of the hangars themselves.

[0003] However, these existing solutions still have the following significant shortcomings:

[0004] 1. Fragmented security management perspective, lacking integrated end-to-end access control. Focusing only on the status monitoring of the drone itself, it fails to treat the drone and the drone as a security collaborative whole that must be jointly verified. It cannot prevent unauthorized drone access or drone impersonation at the access source, resulting in a security blind spot of "emphasizing the drone and neglecting the terminal". This makes the system vulnerable to risks such as impersonation and misuse of permissions during the connection establishment phase.

[0005] 2. The risk identification capability is limited and relies on the trend monitoring of device performance indicators. This single-dimensional analysis is not sensitive enough to complex security threats that require multi-dimensional correlation to identify, such as network intrusion, flight hijacking, and malicious business commands, and is prone to missed detections.

[0006] 3. Weak data-driven and resource adaptive control capabilities, lacking an adaptive computing power allocation mechanism. This results in the inability to guarantee the timeliness of critical security data processing in high-load or unstable link scenarios, or system resources being occupied by low-value data, affecting the real-time performance and efficiency of overall management and control.

[0007] 4. The lack of a proactive maintenance mechanism based on digital twins and predictive algorithms makes it difficult to effectively intervene before hardware failures occur.

[0008] Existing technologies are limited to the state management of the "drone device" itself, failing to construct an integrated, proactive, and closed-loop management system centered on security and covering "drone terminal—intelligent drone nest—business flow—environment." Therefore, there is an urgent need to design a comprehensive security management solution that can span the entire chain from access authentication, operation monitoring, intelligent maintenance to safe decommissioning, thereby systematically addressing the shortcomings of existing technologies. Summary of the Invention

[0009] The technical problem to be solved by this invention is to address the shortcomings of the prior art by providing a method for full life-cycle safety management of intelligent drone nests. This method can perform full life-cycle safety management of intelligent drone nests through a two-way access mechanism, multi-dimensional anomaly detection, predictive maintenance driven by digital twins, and safe retirement processing, which significantly improves the safety, reliability, and maintainability of the nests during use and is applicable to various complex application scenarios such as power line inspection and urban security.

[0010] The solution adopted by this invention to solve its technical problem is as follows:

[0011] A method for full lifecycle management and control of drone nests includes the following steps:

[0012] Step S1: Adopt a two-way access mechanism that combines identity authentication and device fingerprinting, verify the legitimacy of the drone terminal and drone nest through an asymmetric encryption algorithm, and generate access security credentials.

[0013] The device fingerprint includes the UAV hardware model, firmware version, communication module MAC address, sensor calibration parameters, and nest physical identification information.

[0014] Step S2: Integrate multi-source data through edge computing nodes to construct a multi-dimensional visualization relationship model and a dynamic behavior monitoring model;

[0015] Step S3: Complete data classification and screening based on multi-dimensional data quality assessment indicators, realize equipment operation control through adaptive parameter adjustment, and dynamically match computing power allocation by combining data value and resource consumption.

[0016] Step S4: Based on the dynamic behavior monitoring model, an anomaly detection algorithm with multi-dimensional feature fusion is used to identify network intrusion, operational disorder and business parameter fluctuation risks in real time and trigger graded early warnings.

[0017] Step S5: Construct a digital twin of the drone nest equipment. Based on the digital twin of the drone nest equipment, integrate the robot control data, wireless charging data and historical fault data, calculate the remaining lifespan of the drone nest hardware through a lifespan prediction algorithm, and generate an active maintenance strategy. At the same time, use the complete data from fault prediction to maintenance completion as a new training sample. Retrain the lifespan prediction algorithm so that its prediction results become more and more accurate with the accumulation of data, forming a continuously self-optimizing closed-loop prediction model.

[0018] Step S6: Use the AES-256 encryption algorithm to encrypt and erase sensitive data in the drone terminal and drone nest, cancel access credentials, and complete the full lifecycle security loop;

[0019] In steps S1 to S5, a dynamic key update mechanism is used to ensure data transmission security. The key update combines the timestamp and the device trust score to generate a new key. The dynamic key update formula is as follows:

[0020] ;

[0021] in, The updated data encryption key, The old key. This is the current timestamp, accurate to the second. The device trustworthiness score is determined using the SHA-256 hash function. This indicates a string concatenation operation.

[0022] As a preferred embodiment of the present invention

[0023] Step S1 includes the following steps:

[0024] Step S10: Collect the physical identification information of the machine nest using a two-way access mechanism, including the unique code of the machine nest, the motherboard serial number and the location information;

[0025] Step S11: Verify the validity and legality of the digital certificates of the drone terminal and the drone nest using an asymmetric encryption algorithm, and generate access security credentials, including:

[0026] Step S110: The drone terminal receives the digital certificate of the drone's internal storage from the drone's internal storage and parses it, extracting the plaintext portion and the digital signature attached to the digital certificate.

[0027] Step S111: The drone terminal uses the same hash algorithm agreed upon with the issuer to calculate the plaintext part of the digital certificate, obtain a hash value A, and generate a unique "fingerprint" for the certificate content.

[0028] Step S112: The drone terminal uses the pre-stored and trusted issuer root public key to decrypt the digital signature attached to the digital certificate. If the decryption is successful, the original hash value B calculated by the issuer can be obtained.

[0029] Step S12: The drone terminal compares its calculated hash value A with the original hash value B obtained from decrypting the signature.

[0030] If hash value A == hash value B, it proves that the certificate content has not been tampered with during transmission. Combined with the decrypted signature, the original digest is obtained, proving that the certificate was indeed issued by a trusted issuer. Thus, the certificate verification is successful.

[0031] If hash value A ≠ hash value B: This means the certificate content has been tampered with after it was issued; verification fails, and the connection is immediately terminated.

[0032] Step S13: After successful two-way authentication, the drone terminal and the drone nest use a negotiated key or a temporary token issued by the authentication server as the access security credential for this session. This credential is used to encrypt and verify all subsequent communication data to ensure transmission security.

[0033] Step S14: If the verification fails in any of the steps S10 to S13 above, the drone nest will directly reject the access request of the drone terminal and immediately disconnect the connection, while adding the relevant records to the failure log.

[0034] As a preferred embodiment of the present invention

[0035] Step S2 includes the following steps:

[0036] Step S20: Collect UAV communication interaction data, operational data, environmental perception data, and nest hardware status data.

[0037] Step S21: Use Neo4j graph database as storage and computing engine to create entity nodes; establish directed relationships between nodes based on business interaction links, data flow and communication protocols; form a terminal-nest-business association graph, i.e. a multi-dimensional visual relationship model.

[0038] Step S22: Construct a dynamic behavior monitoring model, including:

[0039] Step S220: Collect multiple sets of historical normal operation data to ensure that the data covers various normal business scenarios; perform data cleaning, noise reduction and standardization preprocessing.

[0040] Step S221: Extract multi-dimensional features from the preprocessed data.

[0041] Step S222: Using at least one machine learning algorithm, such as Random Forest or Support Vector Machine, train the dynamic behavior monitoring model to learn the normal correlation patterns and numerical ranges between the multi-dimensional features, thereby establishing a normal behavior baseline.

[0042] Step S223: Use the reserved test dataset to evaluate the dynamic behavior monitoring model. When the dynamic behavior monitoring model achieves an accuracy rate of over 95% in recognizing normal behavior, it is considered to have passed the training. Deploy the trained model to edge computing nodes for real-time comparison of the differences between the current running data and the normal baseline.

[0043] As a preferred embodiment of the present invention

[0044] Step S4 includes the following steps:

[0045] Step S41: The anomaly detection algorithm using multi-dimensional feature fusion extracts communication features, behavioral features, business features, and environmental features.

[0046] Step S42: Calculate the similarity between the current behavior of the UAV terminal and the baseline of normal behavior of the UAV terminal using a weighted fusion formula. The weighted fusion formula is as follows:

[0047] ;

[0048] in, For comprehensive similarity, the value range is [0,1]. The weights for communication characteristics, behavioral characteristics, business characteristics, and environmental characteristics are respectively assigned, satisfying the following conditions: , For communication feature similarity, For behavioral feature similarity, For business feature similarity, For environmental feature similarity;

[0049] Step S43: When the similarity is lower than a preset threshold, it is judged as abnormal, and the following graded warnings are triggered:

[0050] For minor anomalies, a system pop-up notification will appear in the backend management interface to inform maintenance personnel to observe the situation, without taking any proactive intervention at this time.

[0051] In the event of a moderate anomaly, the system automatically sends an SMS notification to the operations and maintenance manager, requesting immediate remote verification and intervention.

[0052] In case of severe anomaly, immediately sever the control and data communication link between the drone terminal and the drone nest, lock the sensitive data stored in the drone nest, prohibit access and transmission, and force the drone terminal to execute an emergency landing or return-to-home procedure.

[0053] As a preferred embodiment of the present invention

[0054] In step S3:

[0055] Data quality assessment indicators include completeness, consistency, timeliness, and relevance.

[0056] Data tiered filtering involves calculating data quality scores using a weighted scoring method and then processing the data according to the quality score values.

[0057] Equipment operation control is based on data quality results, dynamically adjusting sensor acquisition modes, communication protocols, data transmission priorities, and acquisition frequencies.

[0058] The computing power allocation is based on the ratio of data value coefficient to resource consumption coefficient, and the computing power of edge computing nodes is allocated in a differentiated manner.

[0059] As a preferred embodiment of the present invention

[0060] In step S5:

[0061] The lifetime prediction algorithm is a prediction model based on long short-term memory network, which integrates device runtime, load rate, ambient temperature and number of charging cycles to output the remaining hardware lifetime.

[0062] Proactive maintenance strategies should include at least maintenance time windows, spare parts replacement recommendations, and downtime maintenance priorities.

[0063] A full lifecycle management and control system for unmanned aerial vehicle (UAV) nests includes an access authentication module, a data fusion module, a screening and control module, a model building module, an anomaly detection module, a digital twin maintenance module, and a decommissioning management and control module.

[0064] The access authentication module is used to realize two-way identity authentication and device fingerprint verification between the drone terminal and the drone nest, and generate access security credentials.

[0065] The data fusion module is used to collect and integrate multi-dimensional data of UAVs and UAV nests through edge computing nodes, and output the data to the model building module after data cleaning and format conversion.

[0066] The screening and control module is used to complete data hierarchical screening based on multi-dimensional data quality assessment indicators, and to perform adaptive control of equipment operation and dynamic allocation of computing power.

[0067] The model building module is used to build multi-dimensional visual relationship models and dynamic behavior monitoring models;

[0068] The anomaly detection module is used to identify security risks based on a dynamic behavior monitoring model, through a multi-dimensional feature fusion algorithm and a weighted fusion formula, and to trigger graded early warnings.

[0069] The digital twin maintenance module is used to construct a digital twin of the equipment nest and generate proactive maintenance strategies through a life prediction algorithm and a remaining life prediction formula.

[0070] The decommissioning control module is used to encrypt and erase sensitive data, cancel access credentials, and synchronize them to the security management platform.

[0071] As a preferred embodiment of the present invention

[0072] The anomaly detection module includes a dynamic threshold adjustment unit, which is used to optimize the preset threshold in real time based on the drone terminal runtime and changes in business scenarios through reinforcement learning algorithms, thereby reducing the false alarm rate.

[0073] A computer program product includes a computer program that, when executed by a processor, implements the steps of the unmanned aerial vehicle (UAV) nest full lifecycle management method.

[0074] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for full lifecycle management of unmanned aerial vehicle (UAV) nests.

[0075] Compared with the prior art, the present invention has the following beneficial effects:

[0076] 1. This invention breaks through the limitations of existing technologies that only focus on the drone nest itself or provide single-point protection for drones. It treats the drone terminal and the intelligent drone nest as a secure whole that must be jointly authenticated and monitored, spanning the entire process of "access-operation-maintenance-retirement". Through a two-way access mechanism based on device fingerprints and digital certificates, it effectively identifies and rejects identity impersonation and device counterfeiting, preventing illegal devices such as "black flight" drones from accessing the system at the source and improving the basic security of the system. It provides a rigorously verified and trusted data source for subsequent operation monitoring and anomaly detection stages, ensuring that the entire analysis system is built on a reliable foundation. Through a zero-trust strategy of "verification before communication", it completes security confirmation before interaction occurs, transforming passive response into proactive interception, building the first solid line of defense for proactive security defense, and improving the overall security level and reliability of the system.

[0077] 2. Unlike existing technologies that rely on a single performance indicator (such as 4D detection), this invention deeply integrates communication, operation, business, and environmental data, calculates comprehensive similarity through a weighted fusion algorithm, and combines a reinforcement learning-based dynamic threshold adjustment mechanism. This enables the system to have higher sensitivity and accuracy in identifying potential network intrusions, illegal manipulation, and abnormal business behaviors, effectively reducing false alarms and false negatives caused by single-dimensional detection, and ensuring business continuity and security.

[0078] 3. This invention establishes an independent screening and control mechanism, ensuring the quality of data input into subsequent models through multi-dimensional data quality assessment and hierarchical screening. Furthermore, based on the real-time ratio of data value coefficient to resource consumption coefficient, it dynamically allocates computing power to edge computing nodes, achieving "priority processing of high-value data." This ensures the system can guarantee the timeliness of processing critical security data even under high load or complex environments, avoiding resource waste and improving the real-time performance and energy efficiency of overall management and control.

[0079] 4. This invention constructs a high-fidelity digital twin of the equipment nest and uses an LSTM lifetime prediction model to predict remaining lifetime, automatically generating proactive maintenance strategies that include maintenance time windows, spare parts replacement recommendations, and downtime priorities. This transforms post-failure maintenance into predictive maintenance, effectively avoiding business interruptions caused by sudden hardware failures, significantly improving equipment availability and operational economy, while reducing total lifecycle maintenance costs.

[0080] 5. This invention addresses the weakness of existing technologies in controlling the decommissioning process by proposing a military-grade data erasure process of "one-time encryption and two-time overwrite." This completely eliminates the risk of sensitive business data in storage media being recovered and leaked. It also integrates with a security management platform, using a platform-level global blacklist mechanism to globally deregister and register the device's digital identity. Simultaneously, the device status is marked as "decommissioned" and monitoring is stopped in the digital twin system, ensuring the permanent invalidation of the decommissioned device's "digital identity." Even if the device entity is obtained by others, its certificate and fingerprint cannot be used for authentication again, effectively preventing attackers from using the decommissioned device's identity to "disguise" a legitimate terminal and infiltrate the system. This strengthens the security boundary of the entire network, ensuring that the access and decommissioning phases complement each other, forming an impenetrable security loop. This ensures that security control is not phased but permeates every stage from "access" to "operation" to "decommissioning," truly achieving end-to-end full lifecycle security management. This significantly enhances the credibility of the entire UAV nesting system and meets the needs of industry application scenarios with strict requirements for data security and privacy protection.

[0081] This invention constructs a smart drone nest security management system that covers the entire life cycle and has proactive defense and continuous optimization capabilities through collaborative innovation of two-way access mechanism, multi-dimensional anomaly detection, digital twin-driven predictive maintenance, and safe decommissioning. This significantly improves the system's security, reliability, maintainability, and operational efficiency in complex application scenarios. Attached Figure Description

[0082] Figure 1 This is a flowchart illustrating a method for managing the entire lifecycle of a drone nest proposed in this invention.

[0083] Figure 2 This is a schematic diagram of the structure of a drone nest full life cycle management and control system proposed in this invention;

[0084] Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention.

[0085] Explanation of reference numerals in the attached figures:

[0086] 101-Access Authentication Module, 102-Data Fusion Module, 103-Screening and Control Module, 104-Model Building Module, 105-Anomaly Detection Module, 106-Digital Twin Maintenance Module, 107-Decommissioning Management Module. Detailed Implementation

[0087] The specific embodiments of the present invention are described below with reference to the accompanying drawings and examples:

[0088] It should be noted that the structures, colors, proportions, sizes, etc. shown in the accompanying drawings are only used to complement the content disclosed in the specification, so that those skilled in the art can understand and read them, and are not intended to limit the conditions under which the present invention can be implemented. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0089] This invention provides a method for full lifecycle management and control of unmanned aerial vehicle (UAV) nests, comprising the following steps:

[0090] Step S1, Access Phase

[0091] A two-way access mechanism combining identity authentication and device fingerprinting is adopted. The legitimacy of the drone terminal and drone nest is verified using an asymmetric encryption algorithm, and access security credentials are generated. The device fingerprint includes the drone hardware model, firmware version, communication module MAC address, sensor calibration parameters, and nest physical identification information. The process includes the following steps:

[0092] Step S10: Use a two-way access mechanism to collect the physical identification information of the nest, including the unique nest code, motherboard serial number and location information.

[0093] Step S11: Verify the validity and legality of the digital certificates of the drone terminal and the drone nest using an asymmetric encryption algorithm (such as RSA asymmetric encryption algorithm) to generate access security credentials. This specifically includes the following steps:

[0094] Step S110: The drone terminal receives the digital certificate of the drone nest from the drone nest. The digital certificate is issued by the power Internet of Things security management platform and contains device identity information and permission scope. The platform parses the certificate and extracts the plaintext part (holder information, public key, issuer information, etc.) and the digital signature attached to the digital certificate (a digest encrypted by the issuer with its private key).

[0095] Step S111: The drone terminal uses the same hash algorithm agreed upon with the issuer to calculate the plaintext part of the digital certificate, obtaining a hash value A, and generating a unique "fingerprint" for the certificate content.

[0096] Step S112: The drone terminal uses the pre-stored and trusted issuer root public key to decrypt the digital signature attached to the digital certificate. If the decryption is successful, the original hash value B calculated by the issuer can be obtained (the successful decryption itself means that the digital signature was indeed signed by the issuer who has the corresponding private key).

[0097] Step S12: The drone terminal compares its calculated hash value A with the original hash value B obtained from decrypting the signature.

[0098] If hash value A == hash value B, it proves that the certificate content has not been tampered with during transmission. Combined with the decrypted signature, the original digest is obtained, proving that the certificate was indeed issued by a trusted issuer. At this point, the certificate verification is successful, and the drone can trust this certificate and use the public key of the drone's nest contained in the certificate with peace of mind.

[0099] If hash value A ≠ hash value B: This means the certificate content was tampered with after it was issued, verification failed, and the connection was immediately terminated.

[0100] Step S13: After successful two-way authentication, the drone terminal and the drone nest use a negotiated key or a temporary token issued by the authentication server as the access security credential for this session. This credential is used to encrypt and verify all subsequent communication data to ensure transmission security.

[0101] Step S14: If any of the above steps S10 to S13 fail (such as invalid certificate, signature mismatch, device fingerprint mismatch, insufficient permissions), the drone nest will directly reject the access request of the drone terminal and immediately disconnect the connection, and at the same time add the relevant record to the failure log.

[0102] Upon successful two-way authentication during the access phase, a unique, temporary access security credential is immediately generated for the session. This credential replaces any possible static, long-term valid key and is dedicated to encrypting and verifying all subsequent communications of this session, ensuring the confidentiality and integrity of the transmission link from the very beginning of the connection establishment. Furthermore, the two-way admission mechanism implements a zero-trust policy for any failed access attempt, immediately rejecting access and disconnecting the connection. Simultaneously, it mandates the recording of detailed access failure logs, providing first-hand data for security threat analysis, incident tracing, and accountability.

[0103] The two-way admission mechanism in the access phase changes the existing fragile authentication mode that relies solely on digital certificates or static passwords. By strongly binding the digital certificate representing logical identity with the device fingerprint and physical tag representing the physical entity, it achieves dual and cross-verification of "who you are" and "what kind of device you are". At the same time, the access phase also uses asymmetric encryption technology (RSA) to rigorously verify the integrity and authenticity of the digital certificate through a series of cryptographic operations such as certificate parsing, hash calculation, signature decryption and comparison, ensuring that the identities of both communicating parties are endorsed by a trusted third party.

[0104] Step S2, Operation Phase

[0105] By integrating multi-source data through edge computing nodes, a multi-dimensional visualization relationship model and a dynamic behavior monitoring model are constructed, including the following steps:

[0106] Step S20: Collect UAV communication interaction data, business operation data, environmental perception data, and nest hardware status data.

[0107] The data collected includes communication transmission protocol type (e.g., 4G / 5G / Wi-Fi), data packet size and frequency, and communication link status (signal strength, packet loss rate, latency); the data collected includes preset inspection paths, actual drone flight trajectories, control command records, command execution results (success / failure / timeout), and data collection content (e.g., images, videos, temperature readings); the data collected includes temperature, humidity, wind speed, rainfall, and electromagnetic interference intensity; and the data collected includes robotic arm operating parameters (motor torque, motion accuracy), charging module data (output voltage / current, charging efficiency), and cooling system status (fan speed, internal temperature).

[0108] Step S21: Using Neo4j graph database as the storage and computing engine, create entity nodes such as "Drone Terminal," "Smart Dwelling," "Inspection Task," "Data Type," and "Communication Protocol." Establish directed relationships between nodes based on business interaction links, data flow, and communication protocols; forming a terminal-dwelling-business relationship graph, i.e., a multi-dimensional visualized relationship model. Through a graphical interface, managers can view in real time and intuitively: Which drone is performing which task? Which drones are currently housed in a particular dwelling? Which drone generated specific data, and through what link was it transmitted to where? When a node fails, other affected nodes can be quickly located.

[0109] Step S22: Construct a dynamic behavior monitoring model, including:

[0110] Step S220: Collect multiple sets of historical normal operation data (i.e., the four types of data collected in step S20). In this embodiment, more than 1,000 sets of historical normal operation data are collected. Ensure that the data covers various normal business scenarios (such as different weather conditions and different task types); perform data cleaning, noise reduction, and standardization preprocessing.

[0111] Step S221: Extract multi-dimensional features from the preprocessed data, including communication features: average data packet size, variance of communication heartbeat interval; behavioral features: mean and standard deviation of flight speed, average deviation from the predetermined route; business features: mission execution time, image capture success rate; environmental features: average wind speed and temperature range during the mission.

[0112] Step S222: Using at least one machine learning algorithm, such as Random Forest or Support Vector Machine, the prepared normal data features are input into the algorithm to train the dynamic behavior monitoring model to learn the normal correlation patterns and numerical ranges between multi-dimensional features, thereby establishing a normal behavior baseline. This baseline is essentially a multi-dimensional decision boundary. In this embodiment, the Random Forest algorithm, which is suitable for processing high-dimensional features, can evaluate feature importance, is insensitive to outliers, and has good robustness, is preferred.

[0113] Step S223: Use the reserved test dataset to evaluate the dynamic behavior monitoring model. When the dynamic behavior monitoring model achieves an accuracy rate of over 95% in recognizing normal behavior, it is considered to be qualified for training and can be put into actual operation. Deploy the trained model to edge computing nodes for real-time comparison of the differences between the current running data and the normal baseline.

[0114] Step S3, Screening and Regulation Stage

[0115] Data is graded and screened based on multi-dimensional data quality assessment indicators, and equipment operation is controlled through adaptive parameter adjustment. Computing power allocation is dynamically matched by combining data value and resource consumption.

[0116] The screening and control phase serves as a core hub connecting data collection with subsequent anomaly detection and maintenance decisions. Its core objective is to improve data quality and resource utilization efficiency through a closed-loop operation of "data classification - operation control - computing power matching," thereby providing precise data support and stable operation assurance for full lifecycle security management.

[0117] The data quality assessment indicators include completeness (data missing rate ≤5%), consistency (cross-module data error ≤3%), timeliness (data acquisition-transmission delay ≤10s) and relevance (correlation with business / security analysis ≥80%).

[0118] Data grading and filtering involves calculating data quality scores using a weighted scoring method, and then processing the data according to these scores. During calculation, each indicator is first assigned a weight (completeness 0.3, consistency 0.25, timeliness 0.3, relevance 0.15), and then scored from 0 to 100 based on actual performance. The final score is calculated as Σ(indicator score × corresponding weight). Data is then divided into three levels and processed differently: high-scoring data (≥85 points) is directly used for model calculations and real-time monitoring; medium-scoring data (60-84 points) is used after completion and correction; and low-scoring data (<60 points) is marked as invalid, filtered directly, and the reasons are recorded.

[0119] Equipment operation control is based on data quality results, dynamically adjusting sensor acquisition modes, communication protocols, data transmission priorities, and acquisition frequencies. For example, when data quality is ≥85 points, the current sensor acquisition mode (e.g., timed acquisition), communication protocol (e.g., 5G priority), and transmission priority are maintained; when data quality is 60-84 points, the sensor acquisition frequency is adjusted (e.g., increasing from 10s / time to 5s / time), a more stable communication protocol is switched (e.g., 4G backup link), and the data transmission priority is increased; when data quality is <60 points, sensor calibration and communication link detection are triggered, non-critical business data acquisition is suspended, and the focus is on core security data.

[0120] Computing power allocation is based on the ratio of data value coefficient to resource consumption coefficient, with differentiated allocation of edge computing node computing power. In this invention, the data value coefficient (business importance + security relevance) ranges from 0.1 to 1.0, and the resource consumption coefficient (computing power / storage resources required for data processing) also ranges from 0.1 to 1.0. The higher the ratio of data value coefficient to resource consumption coefficient, the higher the data priority. The differentiated allocation method is as follows: edge computing node computing power is allocated hierarchically according to priority. High-ratio data (such as nest hardware status data and UAV control command data) receives ≥60% of computing power resources to ensure real-time processing; medium-ratio data (such as environmental perception data) is allocated 30%-40% of computing power; low-ratio data (such as non-critical business logs) is processed offline in batches, with the remaining computing power allocated. The ratio is recalculated every 5 minutes, and the computing power allocation ratio is adjusted in real time according to changes in business scenarios (such as peak inspection periods or severe weather) to avoid resource waste or delays in critical data processing.

[0121] Step S4: Based on the dynamic behavior monitoring model, an anomaly detection algorithm using multi-dimensional feature fusion is employed to identify network intrusion, operational malfunctions, and business parameter fluctuation risks in real time, triggering tiered early warnings. This includes the following steps:

[0122] Step S41: The anomaly detection algorithm using multi-dimensional feature fusion extracts communication features, behavioral features, business features, and environmental features.

[0123] Step S42: Calculate the similarity S between the current behavior of the drone terminal and the baseline of the normal behavior of the drone terminal using a weighted fusion formula.

[0124] The weighted fusion formula is as follows:

[0125] ;

[0126] in, For comprehensive similarity, the value range is [0,1]. The weights for communication characteristics, behavioral characteristics, business characteristics, and environmental characteristics are respectively assigned, satisfying the following conditions: , For communication feature similarity, For behavioral feature similarity, For business feature similarity, This represents the similarity of environmental features.

[0127] Step S43: Similarity is lower than a preset threshold (S < 0.05) When the condition is found to be abnormal, a tiered warning is triggered.

[0128] For example, preset anomaly detection threshold The initial value is 0.75. When S < 0.75, the following graded warning is triggered:

[0129] When 0.6 ≤ S < 0.75, it is considered a minor anomaly, which may be due to slight fluctuations in individual parameters, such as a slight increase in communication latency. The corresponding warning method is to generate a system pop-up notification in the backend management interface, notifying the operations and maintenance personnel to observe the situation, without taking any proactive intervention at this time.

[0130] When 0.4 ≤ S < 0.6, it is judged as a moderate anomaly, meaning that anomalies occur in multiple dimensions, such as continuous deviation of the drone's flight path and deterioration in the quality of the captured images. The corresponding warning method is that the system automatically sends an SMS notification to the mobile phone of the operation and maintenance manager, requesting immediate remote verification and intervention.

[0131] When S < 0.4, it is considered a severe anomaly, which means that the drone is likely to suffer a network intrusion or a serious failure, such as a sudden change in the communication protocol or an attempt to fly into a no-fly zone. The corresponding warning method is to immediately cut off the control and data communication link between the drone terminal and the drone nest, lock the sensitive data stored in the drone nest, prohibit access and transmission, and force the drone terminal to perform an emergency landing or return-to-home procedure.

[0132] Step S5: Construct a digital twin of the drone nest equipment. Based on this digital twin, integrate robotic arm control data, wireless charging data, and historical fault data. Calculate the remaining lifespan of the drone nest hardware using a lifespan prediction algorithm, generating a proactive maintenance strategy that includes at least maintenance time windows, spare parts replacement recommendations, and downtime maintenance priorities. The lifespan prediction algorithm is a prediction model based on a long short-term memory network, used to integrate equipment runtime, load rate, ambient temperature, and charging cycle count to output the remaining hardware lifespan.

[0133] It also includes using the complete data from fault prediction to maintenance completion as new training samples; retraining the life prediction algorithm so that its prediction results become more and more accurate as data accumulates, forming a closed loop of continuous self-optimization.

[0134] Step S6, Retirement Phase

[0135] Sensitive data on the drone terminal and drone nest is encrypted and erased using the AES-256 encryption algorithm, and access credentials are revoked to complete a full lifecycle security loop. Sensitive data includes inspection images, power equipment parameters, and authentication information, requiring three rounds of overwrite erasure to ensure data is unrecoverable. The steps include:

[0136] 1. Scan all storage media (such as SD cards, internal storage, and solid-state drives) of the drone terminal and drone nest, and identify and classify sensitive data after scanning. Sensitive data types include: business data: inspection images, video recordings, and collected power equipment parameters (temperature, current, voltage, etc.); security data: identity authentication information, historical session keys, access logs, and audit tracks; and configuration data: network configuration, flight path planning, and flight control parameters.

[0137] 2. A random key is generated using the AES-256 algorithm to encrypt the entire storage space or the sector containing sensitive data once, followed immediately by two overwrites of all zeros or random data; that is, a three-stage process of "one encryption and two overwrites".

[0138] Initial encryption: Transforms the original data into unreadable ciphertext;

[0139] First overwrite: Overwrite with all zeros to destroy the ciphertext structure;

[0140] Second overwrite: Overwrite with random data to ensure that the previous magnetic record cannot be detected even by physical techniques.

[0141] 3. After erasure is complete, a random sector is sampled and read to verify that the data has been successfully destroyed and cannot be recovered.

[0142] 4. Access credentials and identity cancellation: On the device, permanently delete or disable the chip area storing the access security credentials; send a device decommissioning instruction to the power IoT security management platform; the power IoT security management platform deletes the device's identity information (digital certificate, device fingerprint) in the central database and revokes all its permission records; add the device identifier to the global blacklist to prevent any future re-access.

[0143] 5. In the asset management system, update the status of the device's unique physical code (such as a QR code or RFID tag) to "retired" and move it from the in-use asset pool to the retired asset pool.

[0144] 6. In the nest digital twin system, locate the corresponding UAV or nest twin; change its operating status from "online", "standby" or "maintenance" to "retired", stop all data collection, monitoring and lifespan prediction calculations for the twin, and archive it to the historical model library.

[0145] At this point, the retired equipment is physically identified and deregistered, and its retirement status is marked in the digital twin of the equipment nest, completing the full life cycle closed loop. This ensures that the equipment can be safely, thoroughly, and irreversibly "removed" from the system at the end of its life cycle. It ensures the complete retirement of the equipment from the aspects of the complete destruction of data assets, the complete deregistration of logical identity, and the termination of the physical entity's state in the digital world, preventing any secondary security risks caused by equipment retirement and achieving a complete security closed loop throughout the entire life cycle.

[0146] In steps S1 to S5, a dynamic key update mechanism is used to ensure data transmission security. The key update combines the timestamp and the device trust score to generate a new key. The dynamic key update formula is:

[0147] ;

[0148] in, The updated data encryption key, The old key. This is the current timestamp, accurate to the second. The device trustworthiness score is determined using the SHA-256 hash function. This indicates a string concatenation operation.

[0149] In this invention, the key update cycle is once every 24 hours by default, which can be configured according to business needs; when C < 60 (low device trustworthiness), an emergency update is triggered, and the update cycle is shortened to 1 hour.

[0150] A whole lifecycle management and control system for unmanned aerial vehicle (UAV) nests includes an access authentication module 101, a data fusion module 102, a screening and control module 103, a model building module 104, an anomaly detection module 105, a digital twin maintenance module 106, and a decommissioning management and control module 107.

[0151] The access authentication module is used to achieve two-way identity authentication and device fingerprint verification between the drone terminal and the drone nest, generate access security credentials, and reject unauthorized device access.

[0152] The data fusion module is used to collect and integrate multi-dimensional data from drones and drone nests through edge computing nodes, and output the data to the model building module after data cleaning and format conversion.

[0153] The screening and control module is used to complete data classification and screening based on multi-dimensional data quality assessment indicators, and to perform adaptive control of equipment operation and dynamic allocation of computing power.

[0154] The model building module is used to build multi-dimensional visualization relationship models and dynamic behavior monitoring models.

[0155] The anomaly detection module is used to identify security risks and trigger tiered warnings based on a dynamic behavior monitoring model, using multi-dimensional feature fusion algorithms and weighted fusion formulas. It includes a dynamic threshold adjustment unit, which optimizes preset thresholds in real time based on drone terminal runtime and changes in business scenarios (such as peak power line inspection periods or severe weather) using reinforcement learning algorithms. This reduces the false alarm rate.

[0156] The digital twin maintenance module is used to construct a digital twin of the equipment in the machine nest and generate proactive maintenance strategies through life prediction algorithms and remaining life prediction formulas.

[0157] The decommissioning control module is used to encrypt and erase sensitive data, cancel access credentials, and synchronize them to the security management platform.

[0158] The system provided by the present invention will be described in detail below through two specific application scenarios.

[0159] Example 1: Implementation of full lifecycle safety management of intelligent drone nests in power inspection scenarios.

[0160] This embodiment is applied to the inspection scenario of 110kV-500kV high-voltage transmission lines, involving 20 inspection drones and 5 smart drone nests, deployed in the coverage area of ​​a provincial power Internet of Things safety management platform. The core objective is to ensure the security of inspection data (equipment defect images, line parameters) and the stable operation of the drone nest hardware.

[0161] Step S1: Access phase,

[0162] Unmanned aerial vehicle (UAV) equipment fingerprint collection: hardware model, firmware version, communication module MAC address, and sensor calibration parameters.

[0163] The data collection includes: unique code, motherboard serial number, and deployment location information; the digital certificate is verified using the RSA-2048 asymmetric encryption algorithm, issued by the power Internet of Things security management platform, which clarifies the inspection authority (limited to designated transmission line sections) and generates an access security credential valid for 7 days.

[0164] When unauthorized devices connect, such as unregistered civilian drones, the system directly rejects the connection, records the log, and pushes it to the power safety monitoring center.

[0165] Step S2, Operation Phase

[0166] During the operation phase, data fusion and model building are carried out. Edge computing nodes collect communication interaction data (4G / 5G transmission protocol, inspection data packet size and link packet loss rate), business operation data (preset inspection path, tower defect identification command execution record and insulator temperature collection data), environmental perception data, and nest hardware status data.

[0167] By using Neo4j to construct a relationship map of "drone-drone nest-transmission line tower", managers can view the historical interaction records between a drone and its nest, as well as the line segments they are responsible for inspecting.

[0168] Input 1200 sets of historical normal inspection data, train the model using the random forest algorithm, and achieve an accuracy of 96.3%. Establish a baseline for normal behavior, with an inspection speed of 3m / s and a data transmission frequency of 5s / time.

[0169] Step S3, the screening and control stage, completes the data classification and screening based on multi-dimensional data quality assessment indicators, realizes equipment operation control through adaptive parameter adjustment, and dynamically matches computing power allocation with data value and resource consumption.

[0170] Step S4: Based on the dynamic behavior monitoring model, an anomaly detection algorithm with multi-dimensional feature fusion is used to identify network intrusion, operational disorder and business parameter fluctuation risks in real time, and trigger graded early warnings.

[0171] Multi-dimensional feature extraction: communication features, behavioral features, business features, and environmental features.

[0172] Weighted fusion calculation: weights The overall similarity S=0.68 is lower than the threshold of 0.75, triggering a mild anomaly warning. The system pop-up window prompts the administrator to check the impact of wind speed.

[0173] Dynamic threshold adjustment: During heavy rain, the threshold is lowered to 0.7 using a reinforcement learning algorithm to improve the sensitivity of anomaly detection in harsh environments;

[0174] Step S5: Based on the digital twin of the UAV nest equipment, calculate the remaining lifespan of the UAV nest hardware using a lifespan prediction algorithm, and generate an active maintenance strategy.

[0175] Based on the Unity3D engine, the hardware such as the machine nest robotic arm, charging module, and lightning protection device are reproduced and the running status is mapped in real time. The LSTM model input data has a running time of 1200h, an average load rate of 72%, an average ambient temperature of 28℃, and 320 charging cycles. The output is a remaining life of 186h, which is 200h below the warning threshold. The system generates maintenance time windows, spare parts replacement suggestions, and downtime maintenance priorities.

[0176] Step S6, Retirement Phase

[0177] The AES-256 algorithm was used to overwrite and erase more than 3,000 inspection images and power line resistance parameters stored in the drone's nest three times. After professional tools detected no data residue, the access security credentials were cancelled, and the data was synchronized to the power Internet of Things security management platform. The device's identity information and inspection permissions were deleted, and the drone and its nest were marked as "retired" in the digital twin system, completing the closed loop of the entire life cycle.

[0178] It enables full lifecycle safety management of drone nests in power grid inspection scenarios, improves the accuracy of anomaly detection, predicts hardware failures in advance, and effectively avoids inspection interruptions and data leaks.

[0179] Example 2: Implementation of full lifecycle safety management of intelligent drone nests in urban security patrol scenarios.

[0180] This embodiment is applied to security patrols in the core business district of a first-tier city, involving 15 security drones and 3 smart drone nests, linked with the city's public security system. The core objective is to ensure the security of public safety data and the stable operation of the drone nests around the clock.

[0181] Step S1, Access Phase

[0182] Two-way access mechanism: Device fingerprints include drone serial number, camera model, and drone nest installation location (such as streetlight pole markings). Digital certificates are issued by the Municipal Public Security Bureau's security management platform, and the key is updated every 24 hours. The device trust score C is calculated based on historical behavior (such as the number of violations), and when the score is below 60, the key update cycle is shortened to 1 hour.

[0183] Step S2, Operation Phase

[0184] Edge computing nodes integrate the following data: communication interaction data: video stream transmission bitrate, Wi-Fi / 5G signal strength; business operation data: patrol routes, facial recognition logs, event response records; environmental perception data: light intensity, crowd density, noise level; and nest hardware status data: number of times the hatch is opened and closed, and battery charge and discharge cycles.

[0185] Dynamic behavior monitoring model: A random forest + support vector machine fusion algorithm is used to train a baseline of normal behavior (such as the drone's cruise altitude mode) and identify abnormal behavior, such as deviation from the predetermined area.

[0186] Step S3, the screening and control stage, completes the data classification and screening based on multi-dimensional data quality assessment indicators, realizes equipment operation control through adaptive parameter adjustment, and dynamically matches computing power allocation with data value and resource consumption.

[0187] Step S4: Based on the dynamic behavior monitoring model, an anomaly detection algorithm with multi-dimensional feature fusion is used to identify network intrusion, operational disorder and business parameter fluctuation risks in real time, and trigger graded early warnings.

[0188] Multi-dimensional feature extraction: Communication features, video stream interruption > 10s; Behavioral features, inspection not following the planned route > 15 minutes; Business features, suspicious person capture missed > 3 times / hour; Environmental features, sudden increase in population density to 80 people / ㎡.

[0189] Weighted fusion calculation: weights If the overall similarity S=0.52 (below the threshold of 0.75), a moderate anomaly warning is triggered, and a text message notification is sent to the public security command center.

[0190] Emergency Response: Management personnel remotely verified that the drone had to avoid the crowd, and after adjusting the inspection route, normal operation was restored. The system recorded the anomaly handling log.

[0191] Step S5: Based on the digital twin of the UAV nest equipment, calculate the remaining lifespan of the UAV nest hardware using a lifespan prediction algorithm, and generate an active maintenance strategy.

[0192] It can map the real-time operating status of the charging module, infrared temperature measurement module, and emergency power supply system, support remote fault simulation diagnosis, input data from LSTM model, run time of 800h, average load rate of 65%, average ambient temperature of 26℃, charging cycle count of 240 times, output remaining life of 235h, no emergency maintenance required, and generate a routine maintenance plan.

[0193] Step S6, the decommissioning phase, involves encrypting and erasing sensitive data, such as patrol videos and facial databases, canceling credentials, and marking the digital twin with a "decommissioned" status.

[0194] It improves the response efficiency of urban security systems, reduces false alarm rates in anomaly detection, decreases maintenance costs, and ensures compliance with public data privacy regulations.

[0195] A computer program product, such as computer program instructions, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the present invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0196] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. The memory and the processor are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to perform the steps of a method for managing the entire lifecycle of a drone nest.

[0197] The preferred embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

[0198] Many other changes and modifications can be made without departing from the concept and scope of this invention. It should be understood that this invention is not limited to the specific embodiments, and the scope of this invention is defined by the appended claims.

Claims

1. A method for managing a full life cycle of a UAV nest, characterized in that, The method comprises the following steps: Step S1, a two-way access mechanism combining identity authentication and device fingerprint is adopted to verify the legitimacy of the unmanned aerial vehicle terminal and the unmanned aerial vehicle nest through an asymmetric encryption algorithm, and to generate an access security credential; The device fingerprint comprises an unmanned aerial vehicle hardware model, a firmware version, a communication module MAC address, sensor calibration parameters, and nest physical identification information, Step S2, multi-source data is integrated through an edge computing node to construct a multi-dimensional visual relationship model and a dynamic behavior monitoring model; Step S3, data is classified and filtered based on multi-dimensional data quality evaluation indexes, device operation regulation is realized through adaptive parameter adjustment, and power distribution is dynamically matched in combination with data value and resource consumption; Step S4, based on the dynamic behavior monitoring model, a multi-dimensional feature fusion anomaly detection algorithm is adopted to identify network intrusion, operation disorder, and business parameter fluctuation risks in real time, and to trigger a hierarchical early warning; Step S5, a digital twin of the unmanned aerial vehicle nest device is constructed, based on the digital twin of the unmanned aerial vehicle nest device, mechanical hand control data, wireless charging data, and historical fault data are fused, the remaining life of the unmanned aerial vehicle nest hardware is calculated through a life prediction algorithm, and an active maintenance strategy is generated; meanwhile, the complete data from fault prediction to maintenance completion this time is taken as a new training sample; the life prediction algorithm is retrained, so that the prediction result becomes more and more accurate with the accumulation of data, forming a closed-loop prediction model that continuously optimizes itself; Step S6, the sensitive data of the unmanned aerial vehicle terminal and the unmanned aerial vehicle nest are encrypted and erased using an AES-256 encryption algorithm, the access credential is cancelled, and a full life cycle security closed loop is completed; In the steps S1 to S5, a dynamic key update mechanism is adopted to ensure data transmission security, a new key is generated in combination with a timestamp and a device trustworthiness score, and the dynamic key update formula is: ； wherein, is the updated data encryption key, is the old key, is the current timestamp, accurate to seconds, is the device trust score, Hash is the SHA-256 hash function, denotes the string concatenation operation.

2. The unmanned aerial vehicle nest full life cycle management method of claim 1, characterized in that The step S1 comprises the following steps: Step S10, the two-way access mechanism is adopted to collect nest physical identification information, including nest unique code, mainboard serial number, and positioning information; Step S11, an asymmetric encryption algorithm is adopted to verify the validity and legitimacy of the digital certificate of the unmanned aerial vehicle terminal and the unmanned aerial vehicle nest, and to generate an access security credential, including: Step S110, the unmanned aerial vehicle terminal receives the digital certificate of the unmanned aerial vehicle nest from the unmanned aerial vehicle nest and analyzes it, extracting the plaintext part and the digital signature attached to the digital certificate, Step S111, the unmanned aerial vehicle terminal uses the same kind of hash algorithm agreed with the issuer to calculate the plaintext part of the digital certificate, obtaining a hash value A, which generates a unique "fingerprint" for the certificate content; Step S112, the unmanned aerial vehicle terminal uses the pre-stored and trusted issuer root public key to decrypt the digital signature attached to the digital certificate, and if the decryption is successful, the original hash value B calculated by the issuer can be obtained; Step S12, the unmanned aerial vehicle terminal compares the hash value A calculated by itself with the original hash value B obtained by decrypting the signature, If Hash Value A == Hash Value B, it proves that the certificate content has not been tampered with during transmission, and the original digest is obtained by decrypting the signature, which proves that the certificate is indeed issued by a trusted issuer, and thus the certificate verification is passed; If Hash Value A!= Hash Value B: it means that the certificate content has been tampered with after issuance; the verification fails, and the connection is immediately terminated; Step S13, after successful two-way authentication, the UAV terminal and the UAV nest use the negotiated key or a temporary token issued by the authentication server as the access security credential for this session, which is used to encrypt and verify all subsequent communication data, ensuring transmission security; Step S14, if the verification fails at any of the steps S10 to S13, the UAV nest will directly reject the access request of the UAV terminal and immediately disconnect, and at the same time, record the relevant access failure log.

3. The UAV nest life cycle management method of claim 1, characterized in that, the step S2 comprises the following steps: Step S20, collect UAV communication interaction data, business operation data, environment perception data and nest hardware state data, Step S21, use Neo4j graph database as storage and computing engine, create entity nodes; according to business interaction link, data flow and communication protocol, establish the directed relationship between nodes; form a terminal-nest-business association graph, that is, a multi-dimensional visual relationship model; Step S22, build a dynamic behavior monitoring model, including: Step S220, collect multiple sets of historical normal operation data to ensure data coverage of various normal business scenarios; clean, denoise and standardize the data for preprocessing; Step S221, extract multi-dimensional features from the preprocessed data, Step S222, use at least one of random forest and support vector machine machine learning algorithms to train the dynamic behavior monitoring model to learn the normal association pattern and value range between the multi-dimensional features, thereby establishing a normal behavior baseline, Step S223, use the reserved test data set to evaluate the dynamic behavior monitoring model, and when the recognition accuracy of the dynamic behavior monitoring model for normal behavior reaches 95% or more, the training is considered qualified, and the trained model is deployed to the edge computing node for real-time comparison of the differences between the current operation data and the normal baseline.

4. The UAV nest life cycle management method of claim 1, characterized in that, the step S4 comprises the following steps: Step S41, the multi-dimensional feature fusion anomaly detection algorithm extracts communication features, behavior features, business features and environment features, Step S42, calculate the similarity between the current behavior of the UAV terminal and the normal behavior baseline of the UAV terminal by a weighted fusion formula, and the weighted fusion formula is: ； wherein, is a comprehensive similarity, and the value range is [0, 1]; respectively are weights of the communication feature, the behavior feature, the service feature and the environment feature, and satisfy , is a communication feature similarity, is a behavior feature similarity, is a service feature similarity, is an environment feature similarity; Step S43, when the similarity is lower than the preset threshold, it is determined as abnormal, and the following graded warning is triggered: mild abnormality, a system pop-up window is generated in the background management interface to notify the operation and maintenance personnel to pay attention to observation, and no active intervention is performed, moderate abnormality, the system automatically sends a short message to the mobile phone of the operation and maintenance person in charge, requiring immediate remote verification and intervention, Severe anomaly, immediately cut off the control and data communication link between the UAV terminal and the UAV nest, lock the sensitive data stored in the UAV nest, prohibit access and transmission, force the UAV terminal to perform emergency landing or return procedure.

5. The UAV nest whole life cycle management method of claim 1, characterized in that, in the step S3, the data quality evaluation indexes include integrity, consistency, timeliness and relevance, the data grading screening is to calculate the data quality score by the weighted scoring method, and the data is processed by grading according to the value of the quality score; the device operation regulation is based on the data quality result to dynamically adjust the sensor acquisition mode, communication protocol, data transmission priority and acquisition frequency; the computing power allocation is differentiated according to the ratio of the data value coefficient to the resource consumption coefficient.

6. The UAV nest whole life cycle management method of claim 1, characterized in that, in the step S5, the life prediction algorithm is a prediction model based on long short-term memory network, which integrates device running time, load rate, environmental temperature and charging cycle number to output the remaining life of hardware; the active maintenance strategy at least includes maintenance time window, spare part replacement suggestion and shutdown maintenance priority.

7. A UAV nest whole life cycle management system adopting the UAV nest whole life cycle management method of any one of claims 1-6, characterized in that, it comprises an access authentication module, a data fusion module, a screening regulation module, a model construction module, an anomaly detection module, a digital twin maintenance module and a retirement management module, the access authentication module is used to realize the two-way identity authentication and device fingerprint verification of the UAV terminal and the UAV nest, and generate access security credentials; the data fusion module is used to collect and integrate multi-dimensional data of the UAV and the UAV nest through the edge computing node, and output to the model construction module after data cleaning and format conversion; the screening regulation module is used to complete data grading screening based on multi-dimensional data quality evaluation indexes, and execute device operation adaptive regulation and computing power dynamic allocation; the model construction module is used to construct multi-dimensional visual relationship model and dynamic behavior monitoring model; the anomaly detection module is used to identify security risks through multi-dimensional feature fusion algorithm and weighted fusion formula based on dynamic behavior monitoring model, and trigger graded early warning; the digital twin maintenance module is used to construct nest device digital twin, and generate active maintenance strategy through life prediction algorithm and remaining life prediction formula; the retirement management module is used to encrypt and erase sensitive data, cancel access credentials, and synchronize to the security management platform.

8. The UAV nest whole life cycle management system of claim 7, characterized in that, the anomaly detection module comprises a dynamic threshold adjustment unit, which is used to real-time optimize the preset threshold by reinforcement learning algorithm based on the running time of the UAV terminal and the change of business scenarios, and reduce the false alarm rate.

9. A computer program product, comprising a computer program, characterized in that The computer program is executed by the processor to implement the steps of the full life cycle management method of the UAV nest according to any one of claims 1-6.

10. A computer device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the steps of the full life cycle management method of the UAV nest according to any one of claims 1-6.

Images