126 results about "Schema analysis" patented technology

The invention discloses an unmanned aerial vehicle monitoring and countering integrated system, and relates to the technical field of unmanned aerial vehicle monitoring, and the system comprises a sensing module which is used for carrying out the monitoring of an environment through a plurality of sensors, and obtaining original data; the fusion module is used for processing the original data by adopting a data fusion algorithm to obtain target information; the identification module is used for outputting the type and behavior mode of the unmanned aerial vehicle; the analysis module is used for evaluating the type and behavior mode of the unmanned aerial vehicle by adopting a dynamic threat evaluation method and generating early warning information; the strategy module is used for formulating a dynamic countering strategy; the execution module is used for implementing corresponding countering measures and monitoring the countering effect in real time; and the recording module is used for recording all results in the whole process. Through the technical means of multi-sensor cooperative monitoring, data fusion processing, classification identification, dynamic threat assessment, game decision making, intelligent countering and the like, omnibearing perception, accurate identification, intelligent assessment and efficient disposal of the unmanned aerial vehicle are realized.

The invention relates to the technical field of artificial intelligence and the field of financial science and technology, and discloses an information processing method based on sentiment analysis, and the method comprises the steps: collecting data and information which comprise text data, user behavior data and event information; performing natural language processing and sentiment analysis on the text data, and identifying sentiment features of the text data; analyzing the user behavior data to determine a behavior pattern; analyzing the event information to determine an event change trend; generating an emotional state index based on the emotional characteristics, the behavior pattern and the event change trend; monitoring the online platform to obtain user emotion feedback; comprehensively analyzing the emotion state index and the user emotion feedback to generate an emotion analysis result; and generating target output content based on the emotion analysis result. Through combination of emotion analysis, behavior data and event change trends, the emotion state of the user can be captured in real time, decision suggestions are dynamically adjusted, and market or environment changes are effectively coped with.

The invention relates to the field of data processing, in particular to a psychological assessment method based on a large model and a related device. The method comprises the following steps: generating a psychological evaluation strategy according to a current interview task and a corresponding psychological evaluation index; generating multi-round dialogue information through an evaluation interaction model based on a psychological evaluation strategy, and interacting with the target object through the multi-round dialogue information to complete a psychological evaluation process corresponding to the current interview task; performing multi-modal interaction mode analysis on the target object based on the multi-round dialogue information and the corresponding response information to obtain a psychological assessment report of the target object; the psychological assessment report at least comprises a quantitative analysis result and a qualitative psychological analysis result aiming at the dialogue content. According to the method, by accurately guiding dialogue content generation and adjusting the evaluation strategy in real time, the psychological state of the target object is more accurately touched, individual differences are adapted, related personnel are assisted in completing psychological evaluation tasks, the accuracy of the evaluation result is improved, and the psychological evaluation efficiency is improved.

The invention discloses a charging pile and system based on power regulation and control. The charging pile comprises a state information acquisition module, a data preprocessing and feature extraction module, a dynamic change mode analysis module, a dynamic change prediction module and a power regulation and control optimization module. The state information acquisition module is used for acquiring each piece of state information of the battery in real time when the charging pile charges the vehicle battery; and the data preprocessing and feature extraction module is used for preprocessing the original data in the data set. By acquiring battery state information in real time, extracting SOC nonlinear change amplitude and thermal management demand change rate, combining data analysis and machine learning prediction, accurately capturing battery charging dynamic change, and dynamically adjusting power regulation frequency by adopting a fuzzy logic algorithm, efficient charging power distribution is realized, the charging efficiency is remarkably improved, and the charging efficiency is improved. Electric energy waste and safety risks are reduced, the service life of the battery is prolonged, the resource utilization rate of the charging pile is optimized, and an innovative technical scheme is provided for intelligent electric vehicle charging.

The invention relates to a data management method and a data management system based on OpenHarmony. The method comprises the following steps: constructing a cross-device data storage layer by utilizing a distributed soft bus technology, and carrying out fragmentation storage and redundant backup on target data to obtain a distributed data storage architecture; performing feature extraction to obtain a data classification result and a fine-grained data mark; comprehensively analyzing the network condition, the equipment performance and the user behavior to obtain a data synchronization strategy; executing two-stage protocol submission and vector clock concurrency control on the cross-device operation to obtain a distributed transaction processing result; performing structural feature and repetitive mode analysis to obtain a compression algorithm selection result and a compression parameter; and carrying out compression processing, and carrying out adaptive routing selection and load balancing on the compressed data transmission path to obtain a multi-device data flow management scheme. According to the method, data integrity and availability are guaranteed, data are effectively compressed, and storage space and transmission bandwidth are saved.

The invention relates to the technical field of cost assessment, in particular to a civil construction project cost assessment optimization system, which comprises a feature analysis module, a cost estimation module, a real-time monitoring module, a model training and tuning module, a risk analysis module, a decision support module, a financial simulation module and a reinforcement learning cost adjustment module. According to the invention, a convolutional neural network algorithm and a principal component analysis method in the feature analysis module excavate building information model data and screen key features, and the cost estimation module combines a multi-layer perceptron neural network and a linear regression model to enhance the cost mode analysis capability and calculate the project cost. The digital twinborn technology and the dynamic programming algorithm realize tracking of project progress and real-time monitoring of cost change in the real-time monitoring module, and the genetic algorithm and the machine learning algorithm realize continuous optimization and self-learning of model parameters in the model training and tuning module. And a Monte Carlo simulation method and a sensitivity analysis method are used for identifying and evaluating project risks in the risk analysis module.

The invention belongs to the technical field of transformer data processing, and particularly relates to a power transformer data cleaning method and system. The method mainly comprises the following steps: obtaining original data of the power transformer, carrying out abnormal data detection and correction on the original data based on physical rule filtering, outlier detection and time sequence mode analysis, extracting different types of data features from the corrected data, and carrying out cross-modal feature fusion on the extracted different types of data features; constructing a fault propagation reasoning initial model according to the data features after cross-modal feature fusion, the basic information of the power transformer and historical test data; and dynamically iterating the fault propagation reasoning initial model through a dynamic ontology updating mechanism to form a fault propagation reasoning model so as to evaluate the state of the power transformer. The space-time barrier of multi-source data is broken through, efficient integration and collaborative analysis of the data are realized, a unified data base is provided for transformer state evaluation, and the utilization rate of the multi-source data is improved.

The invention relates to the technical field of ultrasonic equipment quality control and evaluation, and discloses an ultrasonic quality control dynamic evaluation method and system, and the method comprises the steps: collecting real-time quality control data of ultrasonic equipment based on multi-dimensional feature fusion and dynamic mode analysis, and constructing a quality control database after labeling and preprocessing. The method comprises the following steps: extracting multi-dimensional features from a database to construct a feature matrix, slicing by using an adaptive window segmentation algorithm, and eliminating redundant sub-matrixes by using a dynamic time warping algorithm to form a global feature tensor and similarity association graph; the method comprises the following steps: acquiring current data in real time, converting the current data into a target feature sub-matrix, calculating a dynamic matching degree with a global feature tensor, outputting reconstructed quality control features through multi-modal fusion, and evaluating a quality control state by calculating deviation and combining a dynamic threshold rule. The method and the system can accurately evaluate the quality of the ultrasonic equipment in real time, improve the diagnosis accuracy and have a wide application prospect.

The invention discloses a cross-linking agent production control system and method based on the Internet of Things, and relates to the technical field of chemical production control, the system is composed of a plurality of functional modules, and the system comprises a data acquisition module for obtaining spectral data in a reaction kettle and obtaining temperature data and pressure data of different positions on the wall of the reaction kettle; the data processing and control module is used for converting the spectral data into reaction data of a cross-linking agent intermediate product based on a meta-calibration model; on the basis of the temperature data and the pressure data of different positions, state data, including a local overheating area and a pressure abnormal area, of a cross-linking agent intermediate product are identified by adopting an image identification and mode analysis algorithm; and the decision execution module is used for establishing a dynamic model fusing the production characteristics of the cross-linking agent based on the reaction data and the state data, training an ML model by using historical data and real-time IoT data based on the dynamic model, and automatically generating and executing an optimal control strategy under security constraints.

The invention discloses an autism symptom recognition system based on multivariate emotion path analysis, and the system comprises a multi-modal data collection module which is used for collecting the multi-modal data of a participant, and the multi-modal data comprises skin electrical activity data, eye movement fixation data, electroencephalogram activity data and facial expression data; the data feature extraction module is used for extracting multi-modal features from the multi-modal data; the emotion cognition mode analysis module is used for establishing a path relation model of the multi-modal data and the emotion cognition stage variables, establishing a path relation model of the emotion cognition stage variables and the emotion cognition ability, and calculating to obtain a path coefficient representing each modal; and the deep learning module is used for carrying out orthogonal fusion and autism identification on the mapping features of the multiple modes in an inverted order mode according to the path coefficients corresponding to the multiple modes. According to the method, multi-modal data are systematically collected, orthogonal fusion is carried out on multi-modal mapping features in a reverse order mode, and autism identification can be better carried out.

The invention discloses a power distribution network reliability evaluation method and system based on improved fault mode analysis, and relates to the field of power distribution network evaluation, and the method comprises the steps: obtaining the data information of each contact point in a power transmission line; acquiring power information of each user by analyzing the power consumption data of each user; combining the data information of each contact point in the power transmission line with the power information of each user to obtain an admittance index of each power transmission line; constructing a power distribution network model; establishing a node admittance matrix, and generating a load flow calculation model of the power distribution network; simulating the fault of each contact point in combination with the historical fault data of each contact point and the load flow calculation model; obtaining reliability index change data; and obtaining a fault working index according to the reliability index change data, and evaluating the reliability of the power distribution network. The method has the advantages that the interaction relationship between the constructed element models is fully considered, the accuracy of the fault analysis result is ensured, and the evaluation result is more reasonable.

The invention relates to an intelligent agricultural system based on modular collaboration and AI Agent driving. Environmental sensing, video monitoring and equipment operation data are comprehensively acquired through a data acquisition module; the data processing module carries out real-time abnormity diagnosis and mode analysis; the core framework module cooperatively schedules the data storage module, the user management module and the external interface module to realize cross-layer resource and data integration; the business logic module is connected in series with each layer of events to dynamically generate backlogs and drive multi-domain function execution; and the AI service module carries a multi-modal agricultural agent Agent, has the capabilities of natural language understanding, multi-round dialogue memory, agricultural knowledge reasoning, equipment control decision making, abnormity pre-judgment and personalized suggestion generation core AI, coordinates operation of each module, and realizes intelligent jump from passive response to active suggestion. According to the invention, the problems of response lag, operation splitting and data analysis fragmentation of an existing agricultural system are solved, and the real-time performance, the collaboration and the decision accuracy of agricultural management are remarkably improved.

The invention discloses a patient behavior monitoring method and system based on big data analysis. The big data analysis-based patient behavior monitoring method comprises the following steps: S110, collecting physiological feature data and behavior feature data of a patient; s120, preprocessing the collected physiological feature data and behavior feature data; s130, calculating a physiological feature abnormal value according to the preprocessed physiological feature data; s140, performing behavior pattern analysis according to the preprocessed behavior feature data and the physiological feature abnormal value; through real-time and comprehensive monitoring and analysis, a patient can better manage own health, and the quality and efficiency of medical services are greatly improved on the whole.

The invention provides a geographic information surveying and mapping data intelligent analysis method and system, and relates to the technical field of surveying and mapping data analysis, and the method comprises the steps: determining a to-be-analyzed region, and obtaining geographic information surveying and mapping data; gridding the to-be-analyzed region to form a plurality of spatial positions, constructing a time sequence, and extracting multi-dimensional trend feature vectors in different time windows based on the time sequence; for each spatial position, screening to obtain a final candidate neighborhood; screening to obtain an effective neighborhood according to the multi-dimensional trend feature vector; performing interpolation processing on the time sequence of the target spatial position based on the effective neighborhood to obtain a complete time sequence of each spatial position; and performing spatio-temporal change mode analysis on the complete time sequence, identifying a spatio-temporal evolution rule and generating a spatio-temporal change analysis report. According to the method, accurate identification and intelligent analysis of the spatial-temporal evolution law of the geographic surveying and mapping data can be realized, and an interpretable analysis result with geographic significance is provided.

The invention provides a concrete vibration quality cross-modal detection system. The concrete vibration quality cross-modal detection system comprises a signal acquisition module, a data processing module and a quality analysis module. The data processing method comprises the following steps that a signal collecting module composed of a plurality of acceleration sensors is used for collecting the axial movement acceleration of the vibrating rod and acceleration response data of different positions in the effective vibrating radius; the method comprises the following steps: processing a multi-source sensing signal by adopting a symmetric point mode analysis method, converting a time sequence into a polar coordinate system scatter diagram, and fusing all sensor signals into the same SDP image by adjusting initial parameters of symmetric point mode analysis so as to increase characterization differences of different fault states; performing vibration construction quality analysis on the SDP image by using a convolutional neural network; through manual classification and machine learning training, the CNN model can accurately judge that the vibration construction quality is under vibration, compact or over vibration. According to the scheme, the vibration construction quality can be accurately judged, and the construction quality and efficiency are improved.

The invention relates to the technical field of power-saving control, and discloses an intelligent power-saving control method and system for building construction power utilization, the system comprises a data acquisition unit, a preprocessing unit, a mode analysis unit and a power-saving control unit, the data acquisition unit obtains multi-dimensional target data, builds a foundation for power utilization analysis, and provides a power-saving control system for building construction. The preprocessing unit is used for denoising voltage and current data by adopting a moving average filtering algorithm, calculating parameters of each acquisition point according to a filtering window, providing accurate data for subsequent work, detecting and correcting abnormal values of the voltage, current and power data, judging through a mean value standard deviation and processing according to a mode, so that the accuracy of a flow is enhanced; the mode analysis unit divides intervals according to a power threshold value, counts the operation time and the proportion of equipment, compares the operation time and the proportion with the threshold value, accurately determines a plurality of power utilization modes, and provides a basis for power saving strategy matching; and the power-saving control unit extracts parameters to generate a signal control device according to a power consumption mode matching strategy, thereby realizing an energy-saving target and scientific management and control.

The invention discloses a dynamic micro-expression recognition and psychological anomaly mining system, which comprises a data acquisition module, a multi-dimensional feature extraction module, a feature fusion module, an emotion classification module, a psychological anomaly evaluation module and a feedback and interaction module. According to the system, the accuracy of emotion recognition is greatly improved through multi-modal data collection and feature extraction including micro expressions, voices, limb actions and physiological signals, the psychological abnormal state of a user can be accurately predicted through historical data and behavior pattern analysis, accurate health scoring and early warning are provided, and the health of the user can be accurately evaluated according to an evaluation result. The system can provide personalized emotion regulation suggestions in real time or give out early warning when psychological abnormality is found, and the psychological health management level of the user is improved.

The invention is suitable for the technical field of coating detection, and provides a scraper fineness testing method based on machine vision and an intelligent algorithm, and the method comprises the following steps: capturing a particle distribution image through a 4K line scanning camera and a multi-angle LED light source; through Gaussian-median hybrid filtering, adaptive threshold segmentation (calculating a dynamic threshold based on a neighborhood mean value or a Gaussian weighted mean value), sub-pixel edge refinement (bicubic interpolation) and form / color / texture feature extraction, particles, bubbles, scratches and other interference items are distinguished, A-method and B-method dual-mode analysis is supported, and a particle distribution diagram and a form parameter report are generated. According to the method, manual operation errors are eliminated, the recognition precision of particles with the particle size larger than or equal to 2.5 microns is improved, automation and data traceability of the detection process are achieved, and the method is suitable for paint fineness detection within the range of 0-150 microns.

Systems and methods are provided for multi-schema analysis of patient specific anatomical features from medical images. The system may receive medical images of a patient and metadata associated with the medical images indicative of a selected pathology, and automatically classify the medical images using a segmentation algorithm. The system may use an anatomical landmark detection algorithm leveraging Deep Reinforcement Learning (DRL) techniques to automatically locate one or more anatomical landmarks associated with the patient specific anatomical feature within the medical images. A 3D surface mesh model may be generated representing the patient specific anatomical features including the located one or more anatomical landmarks. The located one or more anatomical landmarks may be used to guide placement of a 3D model of a medical device that may be fused with the 3D surface mesh model to generate a patient specific 3D model of the medical device.

The invention discloses a material-environment coupled self-adaptive riveting parameter optimization method and system. The method comprises the following steps: acquiring force data, speed data, position data, environment data and material data in a riveting process; when the force data is greater than a preset force threshold value, adjusting the parameters of the riveting equipment to obtain stable force data; performing fluctuation mode analysis on the stability strength data to obtain an equipment state report; combining the material data with a preset strength prediction model to obtain initial prediction strength; adjusting the initial prediction force according to the comprehensive correction coefficient to obtain the optimal prediction force; and judging whether the equipment parameters need to be adjusted or not according to the predicted optimal force and stable force data to obtain a judgment result, and adjusting the riveting equipment according to the judgment result and the equipment state report. According to the method, the riveting operation stability and the equipment maintenance efficiency can be improved by predicting and adjusting the riveting force in real time and combining long-term data analysis and identification equipment abrasion and material change.

The invention provides an animal anomaly detection method and system based on a hybrid expert model, and the method comprises the steps: carrying out the modeling of a healthy behavior mode through collecting and analyzing the behavior data of a large number of healthy animals, and extracting a statistical rule of behavior characteristics; the Euclidean distance between the behavior characteristics of the sample to be detected and the health behavior mode is calculated to serve as the basis for judging the abnormal state, image characteristics are considered, and statistical information of the behavior characteristics is integrated; the limitation of the existing animal anomaly detection research on the aspect of judging the anomaly by using apparent data is overcome, and the function of improving the accuracy and reliability of animal behavior anomaly detection is realized. According to the method, the clustering analysis and other methods and genetic algorithms are adopted, behavior factors are comprehensively considered, behavior weights are scientifically distributed, the accuracy and adaptability of behavior pattern analysis are improved, and abnormal behaviors are better captured and recognized; classifying the behavior data by using a plurality of machine learning algorithms, and fusing prediction results of a plurality of models by using integrated learning voting to obtain a final anomaly detection result; advantages of various methods are combined, and accuracy and reliability of behavior pattern analysis are improved.

The invention relates to the field of data processing, in particular to an automatic risk identification method and system for an electric energy meter based on big data of the electric energy meter, and the method comprises the steps: collecting operation data of multiple dimensions of the electric energy meter, segmenting the operation data according to a time window with a fixed time length, and obtaining a data block containing an operation data time sequence of each dimension, performing feature extraction on the current data block and constructing a feature vector; inputting the feature vector into a trained isolated forest model, and calculating an original abnormal score; and correcting the original abnormal score to obtain a corrected abnormal score, constructing an abnormal threshold, and judging that the current data block of the electric energy meter has an abnormal risk when the corrected abnormal score is greater than the abnormal threshold. According to the method, the anomaly detection capability of the isolated forest is deeply fused with physical law and time sequence mode analysis, so that the corrected anomaly score not only contains static feature anomaly information, but also integrates dynamic evolution mode information, and the accuracy and reliability of risk identification of the electric energy meter are comprehensively improved.

This invention discloses a product carbon footprint data management system and method based on the Internet of Things (IoT), relating to the field of environmental protection technology. The method includes: collecting raw data from each stage of a product's entire lifecycle to generate carbon genome metadata; calculating the stage-specific carbon emissions of the product at each stage of its lifecycle based on the carbon genome metadata; constructing a carbon genome sequence of the product based on the carbon genome metadata and the corresponding stage-specific carbon emissions, and performing time-series pattern analysis based on the carbon genome sequence; summarizing the stage-specific carbon emissions of each stage of the lifecycle to calculate the product's total lifecycle carbon footprint; generating a standardized carbon footprint index based on the product's total lifecycle carbon footprint and a dynamic benchmark value; calculating the product's carbon footprint for future periods based on historical carbon genome sequence data; and generating carbon reduction operation parameters based on the product's future carbon footprint prediction results. By accurately collecting and correcting data through IoT sensors, combined with carbon footprint prediction, scientific carbon emission management is provided.

The present invention proposes an analysis method for typical enterprise electricity consumption patterns based on multi-dimensional Isolate-Detect change point detection. First, data cleaning is performed on the collected industrial electricity consumption time series data; then, seasonal decomposition is carried out on the data after data cleaning to eliminate the periodic patterns of the data and extract the residual terms of the data; multi-dimensional change point detection is performed on the residual terms obtained after decomposition based on multi-dimensional Isolate-Detect, and only a single change point is detected in each divided time interval, making the change point detection in the multi-dimensional electricity consumption curve more accurate and the calculation efficiency higher; then, a possible set of change points is obtained by combining the mean test statistic and the maximum test statistic; the multi-sequence fusion is used to screen the possible set of change points for the detected set of change points to determine the final change point positions of the electricity consumption data; finally, based on the obtained change point positions, the standardized time series data is divided to extract the typical electricity consumption behaviors of the electricity consumption.

The invention relates to the technical field of mobile application development, and discloses a mobile application framework rapid construction method and system, and the method comprises the steps: analyzing a state persistence code of a to-be-upgraded basic component, and generating a state mode difference report; recognizing an upper-layer component set based on the component dependency graph and extracting a state access mode; analyzing the state persistence code and the data processing logic of the upper-layer component, and generating a state dependence propagation graph; state change propagation analysis is executed, and an upper layer component state field needing synchronous migration is marked; matching the migration strategy to generate a state migration script and arranging an execution sequence according to an inverse topology sequence; and executing the state transition script to complete component hot replacement and state data loading. According to the method, the problem of cascade state inconsistency caused by non-synchronous migration of an upper-layer component state copy when a basic component state structure is changed is solved.

The present application relates to the technical field of medical image processing, in particular to an ANA cell image multi-mode analysis and prediction method, system, device and medium, through a target detection network, interval cells in an original image are detected, a dynamic sampling mechanism is triggered to cut and sample a sliding window of continuous multiple frames of effective images, and a sample image block set obtained by sampling is input into a constructed multi-label classification network, a CBAM attention module and a category decoupling attention module are embedded in a backbone network of the multi-label classification network, enhanced features are extracted, and the probability that a detection image belongs to one or more categories of cell nucleus types is output. The present application reduces background interference through a dynamic sampling mechanism, enhances small sample category recognition ability through category decoupling attention, optimizes multi-label output through category correlation punishment, and improves the accuracy and clinical practicability of ANA cell image multi-mode analysis.

In embodiments, a method includes dynamically detecting a current revision to a software service running within a container and identifying whether the current revision specifies a significant change or not. In response to the identification, the method further includes selecting a time window in which to analyze the service's usage patterns (the “analyzed time window”) and determining whether to include data regarding the service's usage patterns prior to the current revision in the analyzed time window or not. The method further includes recommending a right-sizing implementation for the service based on the usage patterns in the analyzed time window. In embodiments, until data for the entire analyzed time window has been acquired, the method only implements right-sizing recommendations that increase resources available to the service. Once such data has been acquired, the method implements right-sizing recommendations that both increase and decrease such resources.

The invention discloses a power distribution network architecture evolution planning method, system, equipment and medium, and the method comprises the steps: obtaining power distribution network basic data and social power demand driving factors, and obtaining a power distribution network basic data set through data preprocessing; obtaining a typical development scene set through uncertainty modeling and multi-scene generation and reduction; in combination with system operation constraints, site selection and sizing of the multi-port flexible interconnection device are optimized, and a preliminary site selection and sizing scheme is obtained; a multi-stage investment decision model is constructed and solved, and a preliminary staged construction path is obtained; calculating a power supply reliability index through a fault mode analysis method, and evaluating economy in combination with a cost model to obtain a quantitative evaluation result; and performing global sensitivity analysis through a Morse algorithm, and adjusting an optimization strategy to obtain a final power distribution network architecture evolution planning scheme. The flexibility and the reliability of the power distribution network are remarkably improved.

The invention relates to an intravenous anesthesia instantaneous dose regulation and control method and device based on operation stage recognition, and belongs to the technical field of medical big data and intelligent control. A deep learning model is used for carrying out high-dimensional feature fusion and time sequence mode analysis, current and future surgical operation stages are dynamically and accurately identified and predicted, an ideal target drug concentration is dynamically set by querying a built-in knowledge base generated by big data mining, and a model prediction controller is driven to calculate an optimal instant infusion rate, so that the optimal instant infusion rate is obtained. Advanced and accurate drug regulation and control are achieved, a closed-loop feedback mechanism is introduced, parameters are adaptively adjusted on line according to actual physiological reactions, and control individualization and safety are ensured. According to the invention, the crossing from passive feedback to active predictive control is realized, the stability and safety of anesthesia depth are obviously improved, and the method has great clinical application value.

The invention relates to the technical field of image detection, in particular to a multi-modal data video inspection concentration degree analysis and early warning method and system, and the method comprises the steps: collecting face video data, and extracting eye micro-motion features, eyebrow morphological features and forehead myoelectricity visual features in face features; collecting heart rate variability data and galvanic skin response data in the physiological signals; personalized standardization processing is carried out, and Z-score feature vectors relative to the deviation degree of the user's base line level value are obtained through conversion; on the basis of the Z-score feature vector, constructing a time sequence feature vector, inputting the time sequence feature vector into a multi-modal time sequence fusion network for time sequence collaborative mode analysis, and outputting a current concentration degree probability value and a concentration degree prediction value at a future specified time point; and triggering a grading early warning mechanism based on the current concentration degree probability value and the concentration degree predicted value. According to the invention, information complementation is utilized in multi-modal data fusion, so that the robustness and accuracy of state recognition are improved.