37 results about "Adaptive security" patented technology

The application discloses a kind of anti-network attack's multi-vehicle safety platoon control method and system, the method first establishes multi-vehicle queue dynamics model;Second, design fuzzy state observer to estimate vehicle state and unknown nonlinear dynamics;Then, introduce funnel function to build time-varying performance boundary, and the double constraints of vehicle anti-collision and keep communication are converted into funnel constraints of spacing tracking error;Then, based on backstepping method and funnel performance error, design virtual controller and adaptive law, and obtain adaptive safety controller to obtain control input;Finally, through Lyapunov stability analysis, the tolerance threshold condition that system can still maintain final consistent bounded stable after suffering replay attack is quantitatively obtained.The application simplifies design through funnel control, avoids complex barrier lyapunov function, and for the first time clearly provides quantified anti-interference ability to replay attack, enhances the network security and practicality of multi-vehicle queue system.

The application provides a large language model adaptive security protection method and system, and a storage medium, the method breaks the hysteresis of the traditional defense strategy by constructing a three-agent collaborative architecture of an attack strategy generator, a dynamic defense device and a defense and attack environment evaluator, combining a double-layer optimization and a dynamic evolution mechanism, driving a two-way iteration relying on attack and defense utility quantitative scores, and synchronously evolving the defense capability and the attack evolution; the method constructs a double-layer defense capability for coping with the current situation and predicting the future by real-time adaptation of the defense parameters to the current attack and pre-judgment of the new attack by the defense parameter, improves the generalization to unknown attacks; the method strengthens the learning and identification of high uncertainty attacks by combining meta-learning attack evolution and threat entropy weighted loss, reduces the bypass probability of new attacks; the method balances the security and usability of the large language model by using a dynamic threshold judgment, realizes adaptive and highly reliable security protection in an open scene, and has better generalization.

This application discloses an online detection and rejection system, method, and computer-readable storage medium for automated warehouse pallets. The system includes: a detection module for acquiring three-dimensional geometric data of pallets to be received online; a data processing and judgment module for comparing the acquired three-dimensional geometric data with preset standard pallet data, and determining an abnormal pallet and generating an abnormal signal when the detection deviation exceeds a preset threshold; this module can also evaluate structural integrity based on partition weighting and calculate the center of gravity offset vector; a scheduling system for receiving abnormal signals and locking abnormal pallets; and execution equipment, including automated guided vehicles or sorting mechanisms, for removing abnormal pallets according to the schedule. This application enables high-precision online detection and quantitative damage assessment of incoming pallets, and adaptive safety control of the handling process based on defect data such as center of gravity offset, improving detection accuracy and the overall safety of automated warehousing operations.

The application belongs to the technical field of intelligent power internet of things, and particularly relates to a method for adaptive security access of unmanned aerial vehicle (UAV) heterogeneous terminals for power edge computing, and the specific steps are as follows: terminal registration and trusted anchoring, binding hardware-level identity, collecting multi-modal fingerprints, and constructing a security behavior portrait; adaptive access authentication; dynamic behavior monitoring and baseline adjustment, establishing an identity authentication and management system, synchronously performing multi-dimensional real-time monitoring, linking and disposing abnormal behaviors, and adopting a baseline evolution mechanism to avoid misjudgment; the multi-dimensional real-time monitoring includes terminal business trend monitoring and traffic situation analysis; security closed loop and collaborative response. The application can realize unified risk detection of the UAV, real-time monitoring of vulnerabilities, access permissions and running states of the UAV, construction of a dynamic behavior monitoring model in combination with UAV terminal information, business data, environmental data and the like, and real-time monitoring and early warning of the UAV terminal.

This invention discloses a data security access control method based on symmetric encryption, comprising the following steps: collecting access requests to generate access event sequences; constructing an access behavior graph, binding symmetric key increments, and generating access behavior graph data; calculating multi-scale topological feature encodings into an access behavior fingerprint set, and generating the region to which the access behavior fingerprint belongs; generating an attacker's cognitive state vector; calculating the session information leakage amount and generating an information leakage budget state; selecting a symmetric key subspace and generating a path working key; dividing encrypted data into a semantic field set and generating a candidate access view set; and executing data security access control: when authorized, selecting the authorized view to decrypt and generate the target data access view; when unauthorized, selecting the decoy view to generate the decoy data access view. This invention utilizes access behavior fingerprints and dynamic key control to achieve adaptive secure access control of encrypted data, possessing advantages such as strong anti-inference capabilities, high misleading potential, and refined authorization.

A CPS security assessment and scheduling method based on event-triggered reinforcement learning under time-varying channels is proposed. The method comprises the following steps: (1) establishing a cyber-physical system model and assessment benchmark environment based on time-varying channels; (2) constructing a multi-source assessment utility index and an event-triggered mechanism with hysteresis and refractory period; (3) establishing a partially observable Markov decision process model from the perspective of the assessment end; (4) learning and training assessment and scheduling strategies based on the near-end policy optimization algorithm; and (5) executing event-triggered adaptive security assessment decisions and closed-loop state updates. This invention aims to provide a CPS security assessment and scheduling method based on event-triggered reinforcement learning under time-varying channels that can realize on-demand triggering, adaptive scheduling, and closed-loop assessment of abnormal communication intervention behaviors, and solve the problems of insufficient adaptability to time-varying channels, insufficient characterization of scheduling utility, and weak multi-mode collaborative decision-making capabilities in existing technologies.

The application provides an adaptive safety training and examination method based on a large language model and a vehicle internet, comprising a vehicle-mounted terminal, a cloud platform and a user terminal; comprising: S1: collecting multi-source heterogeneous data and cleaning, and extracting key risk event segments; S2: obtaining driving scene text through driving scene semantic mapping; S3: constructing a regulation knowledge base based on a RAG architecture, and dynamically correlating regulation clause constraints in context through double-path retrieval; S4: generating training content based on a large language model; S5: interactive examination and scoring; S6: closed-loop feedback and model optimization. The application has high pertinence and timeliness, realizes deep semantic understanding and intelligent examination, guarantees regulation compliance, realizes zero illusion, comprehensive safety evaluation, automatic closed-loop management, significantly reduces the labor cost of vehicle fleet management, and improves the efficiency and standardization level of safety management.

This invention discloses a method and system for safety protection of work area boundaries. The method includes: constructing a historical risk heatmap based on historical violation records; generating a real-time motion trend vector field based on real-time motion state data; fusing the two to generate a comprehensive risk field; extracting ridges in the comprehensive risk field whose gradient modulus is greater than a preset risk gradient as candidate risk boundary lines; calculating the rate of change of direction of each point on the candidate risk boundary line, marking high-risk turning points and smooth transition points, and connecting them to form high-risk boundary arc segments and transition boundary arc segments respectively; calculating the risk propagation impedance between the transition boundary arc segment and adjacent high-risk boundary arc segments, and if it is less than a dynamic connectivity threshold, performing arc segment fusion to obtain dynamic risk boundary arc segments; finally, iteratively aggregating all high-risk boundary arc segments and dynamic risk boundary arc segments to generate a dynamic safety protection boundary with dynamic early warning levels. This invention achieves dynamic and adaptive safety protection of work area boundaries.

This invention proposes an adaptive security monitoring method and system based on edge-cloud collaboration and artificial intelligence. The method includes: acquiring monitoring video streams from the edge side and generating a perceived availability state quantity; determining the current security monitoring operation mode based on the perceived availability state quantity at the current moment and the smoothed state quantity stored on the edge side in the previous time slice, through a preset threshold; generating a decision quality summary based on the execution quality statistics results through weighted summation; acquiring and normalizing the perceived stability statistics results to generate a perceived stability direction vector, which is then applied to the construction of the perceived availability state in the next monitoring cycle, forming a continuous adaptive closed loop. This invention can more effectively reduce invalid data interaction and bandwidth consumption, shorten the response path in critical scenarios, improve the long-term stability of the system in complex dynamic environments, and enable the security monitoring system to have stronger adaptive operation capabilities under multi-scenario deployment conditions.

This invention discloses an output compensation control method for multi-agent systems under aperiodic DoS attacks. First, a follower fuzzy approximation system model and an aperiodic DoS attack model are constructed. When the attack activation causes output loss, an attack compensation mechanism is triggered to maintain the last output and provide continuous compensation signals. A switching fuzzy state observer is constructed to estimate the agent states, and an adaptive law for the attack activation and dormancy phases is designed using a Lyapunov function. A globally preset performance function is constructed to establish dynamic performance boundaries, and the consistency tracking error and normalization transformation error are calculated to construct a barrier function. A distributed adaptive safety control law is derived by combining the backstepping method. This invention overcomes the shortcomings of output loss during the attack phase leading to decreased observation performance and the dependence of traditional performance functions on initial error values, ensuring globally preset performance and high-precision state estimation, and solving the cooperative control problem under aperiodic DoS attacks.

The present invention relates to an autonomous agentic artificial intelligence based defense system for protecting critical infrastructure from cyber threats through continuous, validated, and adaptive security operations. The system employs distributed processing units configured to analyze operational telemetry, network communication data, and control signals in real time to identify deviations from established infrastructure behavior. Artificial intelligence based characterization logic is applied to correlate detected deviations with stored threat profiles and infrastructure parameters, followed by multi-stage validation to confirm threat authenticity and severity. Adaptive learning mechanisms dynamically update characterization thresholds and validation parameters based on historical incidents and evolving attack patterns, enabling continuous improvement without manual reconfiguration.

This utility model relates to the field of cable technology, specifically an adaptive security monitoring cable anti-breakage device. The device includes a storage box with a winding roller inside. The surface of the winding roller has a storage groove for collecting the cable. The top and bottom of the storage box are equipped with limiting mechanisms for fixing the winding roller. This utility model, through the winding roller and storage groove, allows excess cable to be hung on a hanging block. A rotating rod drives a rotating column, which in turn rotates the winding roller, thus collecting the cable. A rubber block is then engaged in the arc-shaped limiting groove to secure the excess cable. This effectively collects excess cable, preventing excessive bending and damage from overly long cables, and extending the cable's lifespan.

The application discloses a card reader data transmission encryption method, comprising the following steps: firstly, a collaborative network is constructed in an internal network, a master node is dynamically elected to generate and synchronize a steganography policy configuration file and a hierarchical encryption policy table; when any node reads sensitive data, the master node adopts a secret sharing algorithm to perform fragmentation, sensitivity analysis and grade marking on the original data, and randomly distributes the encrypted fragments; each node independently executes secondary splitting and confusion encryption of differentiated strength according to the fragment grade and the policy table to generate target encrypted data fragments; each node divides the target fragments into micro fragments, dynamically embeds the micro fragments into normal network traffic according to the steganography policy for covert transmission; and the server extracts the micro fragments according to the policy, recombines and decrypts the original data. The application realizes fine-grained adaptive security protection and high concealment transmission, and effectively improves the anti-detection and anti-attack ability of the data transmission process.

The application provides an adaptive security preservation method for a large language model fine-tuning process, and introduces a removable auxiliary security guide mechanism in the fine-tuning training: an initial representation direction for identifying alignment mismatch tendency is obtained based on a small amount of contrast data, in each training round, it is automatically judged which samples are more likely to trigger the alignment mismatch tendency according to the performance of the current batch of samples, and the representation direction is continuously updated, so that it can adaptively adjust following the risk changes in the fine-tuning process. According to the risk strength of the sample, different intensities of intervention are applied to the internal representation of the model, so that the alignment mismatch update is inhibited while the task performance is maintained. After the fine-tuning, the auxiliary mechanism is removed as a whole, so that the final model can still better preserve the original security alignment without relying on additional modules, and a stable balance between security preservation and task utility in the fine-tuning stage is realized.

The application belongs to the technical field of pure hardware physical security and protection, and discloses an integrated pure hardware adaptive security and protection system, which comprises five modules of trend judgment, adaptive power supply, carrier fixation, physical control and hierarchical sound effect output in an integrated packaging shell, each module is composed of discrete components, comparators and logic gate circuits, there is no programmable processor, software program and algorithm, and the modules are only physically linked through hardware wires, level signals and passive logic devices, and the system does not depend on software operation and cloud control. The system realizes anti-disassembly installation through the carrier fixation structure, the adaptive power supply module realizes multi-source automatic switching to ensure uninterrupted power supply, the trend judgment module adopts double trend collaborative logic to reduce false alarms, the physical control module realizes hardware defense, disarm and reset, the hierarchical sound effect output module is suitable for scene output of sound and light vibration alarm, and the system is additionally provided with a passive electromagnetic noise suppression module to improve the anti-interference performance.

The present application relates to the cross field of chip burning and information security encryption, and discloses a chip security burning dynamic encryption verification system and method. The system is composed of seven core modules, and seven core technologies of PUF hardware unique binding, two-way challenge-response authentication, streaming segmented chain encryption, block-by-block real-time verification, six-state adaptive security state machine, key life cycle management and control, and whole-process encryption log tracing are fused. The method is divided into three stages of preparation before burning, dynamic encryption real-time burning and post-burning, realizes one session one key, one chip one binding, whole-process ciphertext transmission and real-time verification, and triggers hierarchical processing and hardware fuse in abnormality. The present application solves the problems of low security, verification lag and key management loss in traditional burning, greatly improves the security, production efficiency and compliance of burning, and is suitable for high-security chip production burning scenes such as automotive, industry and finance.

A computer system and method for managing device trust during digital interactions. The method comprises capturing device data, including static and dynamic parameters, from a device upon user login to a digital platform. Static parameters are analyzed during the initial login to establish a static trust score, while dynamic parameters are analyzed across sessions to establish a dynamic trust score, which is incrementally updated based on subsequent data. The static and dynamic trust scores are combined to generate a combined trust score, and the device is assigned to one of multiple risk clusters. Based on the assigned cluster, adaptive security protocols are triggered. Assignment to a moderate-risk cluster initiates a step-up authentication process, and failure of the process results in assignment to a high-risk cluster, restricting access to the platform.

The application discloses a traffic adaptive safety switching control method and system, computer equipment and a medium, and belongs to the technical field of intelligent traffic, comprising the following steps: constructing a lattice fluid dynamics model, performing linear stability analysis to obtain a critical condition; periodically judging whether the current traffic flow state tends to be unstable based on the critical condition; when the current traffic flow state does not satisfy the critical condition or the traffic oscillation amplitude in the current traffic flow state exceeds a preset threshold, generating an adaptive safety switching controller enabling instruction; activating and running an adaptive safety switching controller based on road network geometric characteristics according to the adaptive safety switching controller enabling instruction; and feeding back the control signal as an additional adjustment term to the flow evolution equation of the lattice fluid dynamics model, so as to obtain a closed-loop control optimized lattice fluid dynamics model. The method improves the anti-interference ability, overall stability and driving smoothness of intelligent networked vehicle traffic flow in a mixed geometric road network environment.

This application relates to an adaptive security incident response method and system based on an AI agent, belonging to the field of cybersecurity. It includes: in the incident perception phase, the AI ​​agent obtains a first incident to be evaluated based on real-time collected multi-source network security data; in the risk assessment phase, the AI ​​agent obtains an attack graph of the network, performs adaptive risk analysis on multiple paths of the first incident to be evaluated within the attack graph, and obtains a secure path for the first incident to be evaluated; in the intelligent decision-making phase, the AI ​​agent generates a response decision for the first incident to be evaluated based on the secure path; and in the response execution phase, the AI ​​agent executes corresponding security protection actions based on the response decision. This application can improve the response efficiency of cybersecurity, reduce reliance on and intervention by humans, promptly perceive security incidents, prevent the expansion of network asset losses, and is applicable to large-scale and diverse cybersecurity protection scenarios, thereby improving the protection capabilities against emerging threats.

The application provides a kind of adaptive security policy generation method and system for zero trust architecture, the initial zero trust risk value is calculated by the physical environment data of the collected internet of things equipment;Based on the application policy adjustment factor corresponding to the sensitive level identifier of the internet of things equipment in the zero trust architecture, the initial zero trust risk value is adjusted to obtain the comprehensive environment risk index;When the comprehensive environment risk index exceeds the first critical value, the resource access control instruction of only opening the basic application program interface is generated, when the comprehensive environment risk index does not exceed the first critical value, the resource access control instruction authorized according to the preset minimum authority range is generated;The resource access control instruction is converted into the adaptive security policy executable by target device;The application realizes the adaptive response of adaptive security policy under zero trust architecture to physical environment change, improves the accuracy, real-time performance and security of adaptive security policy.

This invention relates to a dynamic adaptive security penetration testing method based on reinforcement learning, belonging to the fields of cybersecurity and artificial intelligence. It includes: constructing and dynamically maintaining a dynamic attack graph that represents the target network as belief states; a high-level strategic planner containing a large language model receiving the overall penetration testing task objective and decomposing the objective into a sequence of sub-objectives based on the global state of the dynamic attack graph; for each sub-objective, a low-level tactical planner based on a reinforcement learning agent searching for the optimal action path to achieve the sub-objective on the dynamic attack graph; and an execution module executing this path, with the results fed back in real time to update the dynamic attack graph, enabling the high-level strategic planner to dynamically adjust the plan according to environmental changes. This invention solves the problem that existing automated tools, using linear processes, cannot plan and execute complex attack chains through a hierarchical architecture that combines a large model and reinforcement learning, achieving dynamic adaptation and automatic optimization of attack paths.

The application relates to the technical field of industrial internet, and discloses an industrial internet data security transmission method and application, which solves the technical problems that existing industrial data transmission is not finely classified and encrypted, the encryption mechanism is single, and protection lacks dynamic nature. The method is characterized in that: firstly, industrial data is preprocessed, and classified and graded through a rule engine and a machine learning hybrid algorithm; secondly, a differentiated encryption mechanism is deployed at a transmission layer, a network layer and an application layer, a dynamic trust evaluation model integrating static and dynamic attributes is established for a network node; meanwhile, a lightweight protection module containing identity authentication, access control and anomaly detection is deployed at an edge node, a security monitoring platform is constructed, and encryption algorithms and routing strategies are dynamically adjusted in combination with monitoring data. The application realizes multi-level dynamic adaptive security protection, effectively avoids data leakage and tampering risks, improves production safety stability, takes into account transmission efficiency, and is suitable for intelligent factory industrial internet data communication scenes.

A unified content delivery network system (UCDN) system formed by one or more networks of interoperable peer-to-peer networks. A hierarchical hybrid adaptive secure peer-assisted network system (referred to as SPAN-AI) using a hierarchical AI-driven approach under a unified secure content addressable architecture, the system is based on five key SPAN-AI subsystems: unified naming; unified discovery; hybrid adaptive routing; scalable publish subscribe; and embedded security; all five key SPAN-AI subsystems are securely integrated and jointly optimized via a hierarchical pluggable AI framework with associated simulation, training, and development pipelines embedded in AI agents with varying degrees of awareness and optimization capabilities at the peer-to-peer, edge, core, or other network levels (hierarchy).

Systems and processes enhance cybersecurity by dynamically generating a decoy sub-application that operates parallel to the primary application using generative design and a large language model. The core feature involves real-time anomaly detection within application traffic, utilizing AI to assess threats and orchestrate appropriate responses. Upon identifying potential security threats, the system employs generative design principles to architect a restricted-functionality sub-application, deploying it instantly to engage and analyze the attack vectors without compromising sensitive production data. The sub-application is isolated through software-defined networking, ensuring that its operations do not affect the primary application's functionality. Additionally, sophisticated traffic redirection mechanisms are employed to divert suspicious traffic from the primary to the decoy application, thereby protecting the integrity while allowing detailed threat analysis. This dual-capability system not only safeguards against disruptions but also enhances adaptive security measures through continuous learning and system adjustments.

The disclosed computer-implemented method may include receiving transaction data of an online transaction that has a pending security authentication requirement and determining a risk score and a conversion score from the transaction data. The method may also include assessing combinations of authentication factors for the security authentication requirement based on applying the risk score and the conversion score to determine probabilities of the online transaction being completed with the combinations of the authentication factors. In addition, the method may further include requesting a combination of authentication factors, selected based on the assessing, to satisfy the security authentication requirement for the online transaction. Various other methods, systems, and computer-readable media are also disclosed.

The application discloses a private communication method and system based on a dynamic dictionary and an adaptive security strategy. In view of the problems in the prior art, such as easy identification by an examination system for strong encryption, easy manual discovery for pure camouflage, low matching efficiency of a mobile terminal dictionary, and privacy leakage caused by cloud dependence of intelligent reply, three technical innovations are provided: 1) a low-delay real-time text stream matching of a mobile terminal mass word is realized through a Trie tree, and the time complexity is reduced to O(L); 2) an adaptive security strategy of a camouflage mode and an encryption mode is provided, a user can dynamically switch according to a scene, and the concealment of communication behavior and the absolute safety of content are considered; and 3) a completely localized reply assisting engine, real-time monitoring of trigger words and display of suggested replies, do not upload any user data throughout the process, and completely eliminate privacy leakage from the architecture. The application also supports multi-dictionary set management and encryption synchronization, and is suitable for private communication needs in scenes such as lovers and teams.

This application provides an adaptive security hardening method and apparatus for a power big data sharing transmission channel, relating to the interdisciplinary fields of power system informatization and network security. The method includes: real-time acquisition of operational status information of the power big data sharing transmission channel, including transmission data type information, traffic characteristic information, and external security situation information; inputting the operational status information into a pre-constructed security risk level assessment model, outputting the security risk level of the power big data sharing transmission channel, including low-risk, medium-risk, and high-risk levels; and automatically adjusting the security strategy of the power big data sharing transmission channel according to the security risk level, including encryption algorithm strength, access control granularity, and transmission rate limits. This application can improve the security and transmission efficiency of the power big data sharing transmission channel.