145 results about "Behavioral modeling" patented technology

The invention belongs to the technical field of computers, and particularly relates to an Internet of Things security data real-time monitoring and protection method and system, and the method comprises the steps: collecting equipment communication and state data through an edge agent, and analyzing and extracting standardized metadata; constructing an equipment behavior contour vector based on a sliding window, and dynamically maintaining a global equipment topological graph; triggering a primary alarm in combination with behavior deviation detection and topology abnormity; outputting a threat score and an attack intention through rule matching and Bayesian network double-engine collaborative reasoning; and executing automatic response according to grading, and feeding back and correcting a behavior baseline to realize closed-loop optimization. The system comprises a data acquisition module, a protocol analysis module, a behavior modeling module, a topology maintenance module, an anomaly detection module, a collaborative reasoning module, an automatic response module and a baseline correction module. Through full-link real-time modeling and cross-device collaborative analysis, the attack detection rate is significantly increased to 98% or above, the false alarm rate is lower than 2%, the response delay is controlled within 800 milliseconds, and the security and adaptive ability of the Internet of Things system are enhanced.

The present invention discloses a method and system for artificial intelligence-based cryptocurrency regulatory analysis capable of performing automated, adaptive, and verifiable compliance evaluation across multiple blockchain ecosystems. The invention integrates blockchain data acquisition, data normalization, graph-based behavioral modeling, artificial intelligence inference, and cryptographically anchored reporting within a unified architecture. The system comprises a blockchain data acquisition unit for retrieving multi-chain transaction data, a data normalization unit for harmonizing heterogeneous blockchain formats, a graph construction unit for generating dynamic transaction graphs, a regulatory knowledge base unit storing jurisdiction-specific regulatory rule graphs, an artificial intelligence processor configured for hybrid neural and symbolic reasoning, and a regulatory reporting unit for generating explainable compliance reports cryptographically anchored to a blockchain ledger.

Systems and methods for protecting electronic communications are described. A cyber security appliance may be configured to calculate a topic shift score for a communication by comparing a first lexical profile derived from the communication to a historical lexical profile established for an associated user. This analysis may be performed without using a large language model. The system may also parse communications to extract sensitive data and content from attachments, performing behavioral modeling on the extracted data. Based on the analysis, an autonomous response module may take a variety of mitigation actions. Furthermore, a security mailbox assistant module may perform a secondary, in-depth analysis on user-submitted communications and generate a deterministic report. For outbound communications, a data loss prevention architecture may divert messages for in-line analysis and may include a fail-safe timeout mechanism to ensure service continuity.

The invention discloses a curve driving collision early warning method and system based on multi-risk field fusion, and the method comprises the steps: 1, collecting vehicle dynamics data in real time, and carrying out the short-time prediction; step 2, geometric modeling and rasterization of a road; 3, constructing a geometric risk field; 4, constructing a kinetic energy risk field; 5, constructing a behavior risk field; step 6, constructing a comprehensive risk field; step 7, collision risk judgment; and step 8, generating an active intervention strategy. Collaborative perception and risk prediction of vehicles, roads and driving behaviors are realized through multi-risk field fusion, and potential collision hidden dangers of curve driving are identified in advance; a vehicle dynamics model and risk correction based on road adhesion are introduced to realize accurate risk assessment of the vehicle; multi-time window trajectory prediction and behavior modeling are adopted, so that the recognition capability of complex driving behaviors is improved; a closed-loop system from risk prediction and decision making to active control is realized, and the safety and stability of the vehicle driving in the curve scene are effectively improved.

The invention provides a traffic abnormity and congestion cause analysis method and system based on a knowledge graph, and relates to the technical field of intelligent traffic perception, and the method comprises the steps: carrying out the target detection through employing preprocessed laser radar point cloud data, recognizing traffic participants, carrying out the continuous frame tracking of the traffic participants, and extracting the motion track and behavior characteristics in an event dimension; identifying a behavior event of the traffic target based on the motion trail and the behavior characteristics, binding the identified behavior event of the traffic target to a corresponding target entity, and performing target behavior modeling to form standardized structured information; based on structured information of a traffic target, an intersection-oriented traffic state knowledge graph is constructed, a rule-based abnormal event judgment module is utilized to perform semantic analysis on behavior and event nodes in the traffic state knowledge graph, abnormal traffic events are identified, and potential causes causing traffic congestion are traced. According to the invention, refined understanding and active perception of the intersection traffic state can be realized.

The invention discloses an enterprise carbon emission measuring and calculating method and system based on production activity carbon footprint analysis, relates to the technical field of enterprise carbon emission, and effectively solves the problem of difficulty in data space-time alignment and process association in a complex production process by introducing a time sequence mapping and process feature decoupling mechanism. Multi-source heterogeneous data are fused in the feature recognition stage, and the integrity and interpretability of energy consumption behavior modeling are improved by combining graph structure modeling and causal attribution analysis methods; structured expression and dynamic weight updating of a carbon emission path are realized based on a carbon emission calculation map, and the response capability and traceability of the model to working condition changes are enhanced; high efficiency and adaptability of carbon emission measurement and calculation are realized through distributed path analysis and a carbon factor dynamic adjustment mechanism; the finally output time-phased and process-divided carbon emission result provides a scientific basis and technical support for an enterprise to carry out refined carbon performance evaluation, carbon asset management and green transformation decision-making.

The invention provides a user behavior tracking and portrait generation system, which is characterized in that multi-source original behavior data is acquired through a data acquisition module, and a standard behavior event stream is generated through cleaning and standardization; the behavior modeling module is used for carrying out cross-event association and self-defined time window combination modeling on a standard event flow based on a configurable rule engine to generate a complex event flow; the real-time processing and calculation module performs real-time aggregation calculation on the complex event stream, generates a real-time behavior index, a trigger signal and an incremental portrait snapshot in combination with a preset rule, and pushes a behavior trigger signal to a downstream service system; the user portrait management module dynamically updates user tag weights and values according to the real-time indexes, and generates a target user portrait tag library and a lightweight tag change event stream; the data storage and query module stores data of each link; and the visual configuration and operation and maintenance module issues a configuration instruction through a visual interface, and monitors full-link operation and task scheduling. And real-time accurate portrait construction and instant service response are realized.

The invention discloses an abnormal behavior detection method and device, a nonvolatile storage medium and electronic equipment. The method comprises the following steps: acquiring user behavior data; behavior feature vectors are determined in the user behavior data, and the behavior feature vectors comprise at least one of a mouse behavior feature vector, a keyboard behavior feature vector, a process behavior feature vector and a window interaction feature vector; and analyzing the behavior feature vector by using a machine learning model to obtain an anomaly score output by the machine learning model, the anomaly score being used for representing the degree of deviation of the behavior feature vector from a preset normal behavior feature vector. According to the method and the device, the technical problem that the accuracy of distinguishing the normal user activity from the remote control Trojan abnormal operation is insufficient due to the fact that a related abnormal behavior detection method lacks a refined behavior modeling technology and a dynamic behavior analysis mechanism is solved.

The invention relates to the technical field of occupational simulation training evaluation, and discloses a system for college student occupational scene simulation training evaluation. The system comprises a task analysis unit, a dynamic behavior modeling unit, a multi-mode interaction management unit, a state evolution tracking unit, a strategy adaptation unit and a comprehensive evaluation generation unit. The system receives the initial scene description and generates a structured task framework, and behavior parameters are configured for student roles; constructing a multi-modal interaction atlas by fusing aligned voice, action and sight line data in real time; identifying a key decision node, and dynamically tracking a team state evolution path; matching and adjusting strategies in a behavior rule base according to the evolution sequence, and outputting an adaptive intervention scheme; and finally generating a multi-dimensional structured evaluation report. According to the system, deep insight and dynamic and objective intelligent evaluation of complex interaction behaviors in the simulation training process are realized.

Techniques for performing prediction based on relative position data and / or absolute yaw data are described herein. A vehicle may detect an object in an environment. The vehicle may generate an embedding associated with the object and input the embedding into a machine learned model. The machine learned model may add the absolute yaw of the object to the embedding, generate a rotation matrix based on the pose of the object, and apply the rotation matrix to the embedding. Based on modifying the embedding (or generating a modified embedding), the attention layer of the machine learned model may perform attention on the modified embedding and respond by outputting output data. The vehicle may rotate the output data (or generate rotated output data) which may be used by one or more machine learned models to predict object behavior, generate vehicle actions, etc.

The invention relates to the technical field of behavior modeling, in particular to a multi-scene market declaration and execution coupling method based on energy storage behavior response modeling, which comprises the following steps of: establishing an index, confirming continuity and boundary connection of a declaration direction, and introducing structured jump pairing logic on the basis of constructing a cross-market behavior sequence; a high-frequency behavior chain in a time sequence is accurately captured, the network attribute of a jump structure is remodeled, the structure recognition capability between behavior segments is enhanced, a dynamic transition rule of the behavior sequence is established by introducing time window recognition and tail segment anomaly positioning operation between the jump segments, continuous interference caused by data abrupt change or delay jump segments is effectively eliminated, and the stability of the behavior sequence is improved. Through multiple innovative actions of sequence construction, behavior clustering, structure learning, time sequence comparison and price intersection pairing, a full-chain closed loop from response to declaration behavior correction is formed, and multiple improvements of market behavior understanding ability, data jump expression precision and real-time correction response ability are realized.

The invention relates to the technical field of data processing. The invention provides a back-end interface anomaly detection system based on an access behavior mode, and the system comprises a data collection module which collects interface access logs and behavior data of different data sources, carries out the preprocessing, and transmits the interface access logs and behavior data to a behavior modeling module; the behavior modeling module is used for grouping user types, generating a normal behavior baseline through unsupervised modeling, and caching the normal behavior baseline to a model library; the anomaly detection and analysis module is used for comparing the real-time access data with a normal behavior baseline in a model library, calculating a behavior deviation and an anomaly score and sending the behavior deviation and the anomaly score to the alarm and response module; the alarm and response module is used for judging whether abnormity occurs or not according to a set threshold value and rule, triggering a response mechanism of a corresponding level and generating an alarm event and disposal feedback; and the visualization and feedback module is used for displaying the alarm event and the disposal feedback on a monitoring platform and carrying out manual verification.

The invention relates to the technical field of thermoelectric Internet of Things security, in particular to a thermoelectric Internet of Things terminal data credible situation awareness and protection platform, which comprises an integrated architecture for constructing a distributed probe acquisition layer, a behavior modeling and knowledge base, a dynamic trust evaluation engine, a situation awareness and decision center and a credible traceability module. Multi-dimensional trust evaluation is realized by fusing terminal behavior consistency analysis, service logic compliance verification and group behavior correlation detection, a self-adaptive weight adjustment mechanism is introduced to adapt to scene change, a hierarchical protection strategy is executed based on a dynamic credibility score, and non-interruption of network disconnection protection is guaranteed by relying on edge calculation. According to the method, the data credible sensing precision is improved, dynamic differential protection is realized, the operation and maintenance cost is reduced, multiple thermoelectric scenes are adapted, and a safety guarantee is provided for stable operation of a thermoelectric Internet of Things system.

The invention discloses a network information security access control system based on dynamic trust evaluation, and the system comprises a behavior collection module which collects original behavior log data, and builds a behavior feature sequence reflecting an entity security state through a preset streaming processing engine; the behavior analysis module is used for carrying out long and short term behavior modeling analysis on the behavior feature sequence to obtain long and short term behavior pattern representation, outputting potential causal chain data based on the long and short term behavior pattern representation and an abnormal causal inference model combined with context perception, and constructing a dynamic trust portrait according to the potential causal chain data; the long and short-term behavior modeling analysis comprises long behavior modeling analysis and short behavior modeling analysis; and the strategy generation module is used for driving a strategy engine in real time based on the dynamic trust portrait, carrying out attribute extraction and dynamic attribute coding based on a behavior sequence in a software defined boundary frame, and dynamically generating and issuing an optimal dynamic access strategy with attributes according to a minimum permission principle.

The invention relates to the technical field of computer hardware, and particularly discloses an intelligent control system for a desktop computer mainboard on-board BGA central processing unit DDR6 memory structure, which comprises a hardware sensing layer, a behavior modeling layer, a collaborative decision-making layer and a dynamic execution layer, intelligent cooperative control of a processor and a memory subsystem from physical and electrical characteristics to logic access behaviors is realized by sensing a hardware state in real time, constructing a digital twin model, collaboratively optimizing a prefetching and scheduling strategy and executing feedback in a closed loop, so that the memory access efficiency is improved, the delay is reduced, and the system stability is enhanced. The system not only pays attention to an access mode of a software level, but also quantifies physical layer constraints such as signal integrity and power supply noise into computable model parameters, so that a subsequent optimization decision is established on the basis of digital twinning reflecting a real hardware state, and the problem that traditional control logic is disjointed from physical reality is solved; and a data foundation is laid for unification of high performance and high stability.

The invention discloses a personalized training resource recommendation method based on a graph fusion neural network, and the method comprises the steps: obtaining text data information, capturing the global and local information of a text through a convolutional neural network, adjusting a mapping function according to a constraint condition, and obtaining a multi-modal data model; optimizing the performance of the multi-modal data model through cross compression, improving the multi-modal data model by using a loss function, extracting data information of the multi-modal data model, and obtaining a training resource knowledge graph through data fusion; the method comprises the following steps: acquiring user online behavior data, performing modeling and group clustering on user online behaviors, and performing deep knowledge tracking by utilizing CRU to obtain a learner knowledge evaluation and deep learning model; through a graph fusion neural network recommendation algorithm, the training resource knowledge graph is matched with personalized demands of the user, personalized resource recommendation is obtained, richer semantic information is provided, and the recommendation algorithm can more accurately understand the demands and interests of the user.

The invention discloses a load-side resource demand response scheduling method, system, equipment and medium under a master-slave game, and relates to the technical field of power system scheduling and demand side management, and the method comprises the steps: collecting the normal operation data of a park, generating a load and photovoltaic curve, analyzing and dividing price-type and excitation-type loads, and constructing a model for describing the dynamic response of the loads. The method comprises the following steps: establishing an operator-dominated and load-followed master-slave game double-layer optimization model, inputting unit price, carrying out iterative solution to obtain an optimal price, implementing demand response, calculating energy purchase cost, solving lower-layer optimization based on actual data and cost, and determining energy storage output power. According to the method, a complete technical chain from load classification, behavior modeling, master-slave game optimization to hierarchical solution is constructed, so that accurate perception, behavior prediction and cooperative scheduling of multi-load-side resources are realized, the interests of multiple subjects are effectively balanced on the premise of ensuring safe operation of a power grid, and the flexibility of the power distribution network is remarkably improved.

The invention discloses a data flow security control method and system based on dynamic trust evaluation, and belongs to the technical field of data security, and the method comprises the steps: obtaining a behavior feature sequence of a data request entity in a fixed time window; constructing a behavior map to identify a typical behavior mode; calculating a risk weight under the context by combining the network topology change, the data sensitivity and the upstream and downstream entity states; inputting the behavior map and the risk weight into a historical trust evolution model to generate a current trust value; matching an access control strategy according to the trust value and the data sensitivity; access operation is executed, and behaviors and results are fed back to the trust model for self-adaptive updating; by introducing a behavior modeling and feedback learning mechanism, dynamic regulation and control and real-time risk response of the data access permission are realized, and the security and intelligence in the cross-domain data circulation process are improved.

The invention relates to a multi-agent collaborative intelligent guiding method for an urban rail transit station yard, which belongs to the technical field of intelligent traffic and comprises the following steps: S1, fusing multi-source heterogeneous sensing data into a dynamic digital twinborn body with behavior modeling capability; s2, a simulation optimization closed loop and game mechanism is introduced, a system regulation and control problem is converted into a collaborative excitation problem for the intelligent agents, and collaborative optimization regulation and control of the multiple intelligent agents are achieved; and S3, constructing a context-aware multi-mode man-machine interaction system, and realizing the transmission of a guidance strategy.

The invention provides a loading and unloading progress monitoring method and system based on LLM model information analysis, and relates to the technical field of information analys.The method includes the steps that semantic analysis and role behavior modeling are conducted on loading and unloading related communication content, and task site photos and text statement vectors are fused to form joint vectors; secondly, extracting task state nodes and change paths thereof under the participation of multiple parties, and performing time positioning on task states based on time clues in a context by combining with records of external equipment, so as to form a state-time evolution track of the whole process of the task; according to the method, a set of modeling framework which can be aligned with actual records of the system and identify task deviation is constructed, and the method can be used for intelligent management scenes such as delay early warning, anomaly analysis and progress visualization. The scheme has the advantages of being high in engineering applicability, low in deployment cost, extensible in model and the like, and is particularly suitable for medium and large logistics enterprise loading and unloading business scenes with complex multi-role scheduling, frequent manual interaction and incomplete data structuring.

The invention provides an energy management dynamic planning method for a light storage system under time-of-use electricity price, and the method comprises the steps: building a user behavior feature matrix through collecting and normalizing the start-stop sequence, power change, frequency and other multi-dimensional behavior data of electric equipment, quantifying the uncertainty of a user load through a Bayesian inference and sequential Monte Carlo method, and carrying out the optimization of the user behavior feature matrix. The calculation of the regional level load uncertainty index is realized in combination with a weighted variable coefficient; based on an S-shaped response curve and a moving average mechanism, dynamically adjusting the weights of economical and stable targets, driving multi-target dynamic planning to solve an energy scheduling strategy, and considering both the economical efficiency and the scheduling stability; user behavior modeling and parameter adaptive adjustment are optimized through closed-loop residual analysis feedback, and the accuracy of the model for dealing with load fluctuation and equipment state change and the real-time performance and robustness of scheduling optimization are greatly improved.

A flexible production line and mixed line production process modeling and anomaly prediction method relates to the field of production line anomaly prediction, and comprises the following steps: S1, constructing a final flexible production line operation simulation model driven by real-time monitoring data; s2, constructing an anomaly detection model of the monitoring data of the whole manufacturing cycle of the product; and S3, integrating the flexible production line operation simulation model established in the S1 and the anomaly detection model established in the S2. Operation behavior modeling of the flexible reentrant production line and whole production process anomaly prediction of multi-variety product mixed line production are achieved, support can be provided for manufacturing process parameter optimization, and the product percent of pass and reliability are improved.

The invention is suitable for the technical field of artificial intelligence, and provides a behavior prediction method and device based on time sequence perception and a graph attention mechanism. According to the behavior prediction method based on the time sequence perception and the graph attention mechanism, the recent weight is calculated based on the timestamp difference value, embedding is dynamically adjusted, the attenuation effect of user interest can be effectively captured, the model preferentially pays attention to recent interaction behaviors, and therefore the response speed and accuracy of short-term preference changes of the user are remarkably improved. And secondly, through a graph attention mechanism, explicitly modeling structural dependency (such as a conversion relation from browsing to purchasing) among multiple behaviors, the limitation of traditional single behavior modeling is overcome, and the semantic richness of project representation is enhanced. And a multi-head self-attention module is used for carrying out sequence modeling on structure enhancement embedding, so that a long-term behavior dependency relationship is accurately captured, the defect that a graph structure is insufficient in continuous description of a time sequence is overcome, and complementation of the time sequence and structure information is formed.

This invention provides a theory-data dual-driven method for modeling vehicle car-following behavior. The method includes: collecting a dataset of car-following trajectories from manually driven vehicles; calibrating a 2D-IDM model using the car-following trajectory dataset; using the 2D-IDM model to predict the speed of the vehicle convoy to obtain simulation data; building a KAN4CF deep learning model; pre-training the KAN4CF deep learning model using simulation data and then fine-tuning it with real data to obtain a trained KAN4CF deep learning model; using the speed difference, position difference, vehicle speed, and preceding vehicle speed at each time step within a predetermined time period as input data; extracting features from the input data using a row-cross attention mechanism; and then inputting the extracted features into the trained KAN4CF deep learning model; the trained KAN4CF deep learning model outputs the vehicle's acceleration value at the next time step. This invention introduces randomness into the deep learning model, ensuring that the model achieves high short-term prediction accuracy while the traffic flow simulation results accurately reflect reality.

This invention relates to a deep learning-based autonomous obstacle avoidance and path planning system and method for unmanned aerial vehicles (UAVs). The system includes: a perception simulation module, which uses multi-sensor joint modeling to achieve dynamic environmental reconstruction; a decision control and dynamics modeling module, which generates a series of control commands after acquiring fused perception data; and a scene interaction module, which realizes dynamic obstacle behavior modeling and multi-scene library access. The perception simulation module, decision control and dynamics modeling module, and scene interaction module are integrated with a ROS-MATLAB joint interface platform to achieve real-time synchronization and visualization analysis of multi-source heterogeneous data. This invention proposes a dynamic obstacle prediction model driven by multimodal data, using attention weights to dynamically allocate sensor confidence, reducing obstacle velocity prediction error to 0.12 m / s; and designs a hierarchical reward function and a priority experience replay strategy, improving the DRL training convergence speed to 40% and achieving a Pareto optimal state for path length and energy consumption.

The invention discloses an air-ground cooperative group target formation identification method based on learnable prior, and the method comprises the following steps: 1, generating behavior prior data, and constructing a dynamic group target behavior prior knowledge base in combination with an improved multi-agent depth deterministic strategy gradient algorithm; step 2, ground end group target formation form classification identification and prior feature extraction; and step 3, taking the prior feature vector group as input, and adopting an improved D-S theory to complete no-load end formation identification. In order to give consideration to the real-time processing capability and the recognition precision of a no-load platform, the method adopts an air-ground cooperative processing architecture, completes high-complexity depth feature extraction and behavior modeling at a ground end, obtains prior information and transmits the prior information to a no-load end, realizes quick response through fusion reasoning, and combines an improved D-S theoretical framework to fuse multi-source prior data. The method can effectively improve the recognition robustness and reasoning credibility of a no-load end in a high-conflict environment.

This invention relates to the interdisciplinary field of network security and artificial intelligence, and discloses a method for anomaly mining and sample generation in encrypted traffic. The method includes: data collection and preprocessing to construct a dataset of normal encrypted traffic under multiple scenarios and a dataset of a small number of real malicious samples; deep behavioral modeling of normal encrypted traffic under multiple scenarios based on a large model to construct a baseline security behavior knowledge base; fine-grained comparison of the features of the encrypted traffic to be detected with normal traffic based on an improved contrastive learning algorithm to mine hidden anomaly features; generating encrypted malicious samples based on a controllable generation architecture of the large model, combining hidden anomaly features and known attack patterns; optimizing the encrypted traffic anomaly detection model using the generated encrypted malicious samples, and feeding the optimized encrypted traffic anomaly detection model back to the anomaly feature mining stage to form a closed-loop iterative optimization system. Using this invention, the accuracy, generalization, and real-time performance of encrypted traffic anomaly detection can be improved.

The application relates to the technical field of industrial control safety, in particular to a PLC runtime abnormal behavior identification method and system based on a behavior model, which comprises the following steps: S1, in the PLC program running process, a behavior monitoring module performs real-time monitoring and preliminarily judges abnormal behavior; S2, normal behavior and abnormal behavior are filtered out respectively according to the monitoring results of the behavior monitoring module, and behavior rule extraction is performed on the behavior monitoring module; and S3, behavior modeling is performed based on the extracted behavior rules, normal behavior model generation or update of behavior rule information files is performed on the detected normal behavior, and abnormal behavior model learning of abnormal characteristics, rapid abnormal identification and measure processing are performed on the detected abnormal behavior. The application uses a data model to describe the normal behavior model of the control behavior and the process behavior through the dependent relationship of the normal input and output data of the control and controlled process, so that the intrusion behavior can be accurately identified, and active defense can be performed.

The invention discloses a period precise simulator design method for a combinable core particle network, and relates to the technical field of core simulation testing, and the method comprises the steps: designing a simulation frame of a combinable core particle network simulator; wherein the simulation framework comprises a two-stage core particle network configuration unit and a multi-thread parallel simulation framework; carrying out combinable topology modeling, modular routing mechanism modeling and heterogeneous router micro-architecture modeling on the combinable core particle network; performing protocol layer protocol conversion modeling, protocol layer flow control mechanism modeling, adaptation layer retransmission mechanism modeling and physical layer electrical behavior modeling on the core particle interconnection protocol interface; based on the above design, the period precision simulator for the combinable core particle network is constructed. By designing the simulation framework, more accurate actual core particle network behaviors can be obtained; a higher simulation speed is realized through a multi-thread parallel simulation framework, multi-thread parallel accelerated simulation under a large-scale network is supported, and the simulation test efficiency is improved.

The invention discloses a food detection risk assessment method and system based on artificial intelligence, and relates to the technical field of intelligent food detection, and the method comprises the steps: collecting food attribute data, carrying out the behavior modeling of dynamic microorganisms, and extracting the correlation characteristics of the behaviors of the microorganisms and environment variables; and a grey correlation analysis method is used to calculate correlation weights of the correlation features in different situations, a PCA method is used to optimize the correlation features and obtain a feature matrix, and a multi-model fusion method is combined with a meta-learning algorithm to construct a food risk assessment model and output a risk assessment result. Through combination of key feature weighted optimization and multi-model fusion prediction, the accuracy and stability of food microorganism risk assessment are improved, the adaptability of a food risk assessment model to a complex environment is enhanced, dynamic optimization of a risk assessment result is realized, and the problems of low assessment precision and poor universality of an existing method are solved.