323 results about "Recurrent neural nets" patented technology

The invention relates to the technical field of battery thermal management, in particular to a battery thermal management control method and system, and the method comprises the steps: obtaining the basic operation parameter data of an automobile battery in real time, and carrying out the multi-dimensional data fusion, and obtaining a multi-dimensional data set; based on the thermodynamic model and the multi-dimensional data set, the temperature distribution condition in the battery is calculated in real time, and real-time temperature distribution information is obtained; based on the historical data and the operation state, the future temperature change trend of the battery is predicted through a recurrent neural network model; dynamically adjusting a control strategy of the battery thermal management system according to the real-time temperature distribution information and the prediction result; and analyzing the temperature difference condition of the battery monomers, detecting abnormal temperature rise, and performing temperature difference balance adjustment and triggering a multi-stage protection mechanism when the abnormal temperature rise exceeds a threshold value. According to the invention, the problems of inaccurate temperature prediction and response lag in the prior art are effectively solved, the control strategy is dynamically adjusted through real-time data fusion and recurrent neural network prediction, temperature difference balance and abnormal temperature rise detection are realized, the safety of the battery is improved, and the service life of the battery is prolonged.

The invention relates to a multi-objective optimization scheduling method and system for a multi-AGV flexible job shop. The method comprises the steps of collecting job shop layout data, AGV operation parameters and production task data, and constructing a scheduling system basic model; performing recurrent neural network training and physical constraint model integration to obtain a hybrid scheduling prediction model; subtask decomposition and path search processing are carried out, and a dual-objective optimization model is obtained through constraint linearization conversion; performing particle swarm optimization solution on the dual-objective optimization model, and obtaining a non-dominated solution set through two-stage iterative calculation of an adaptive weight strategy and constraint conversion; performing multi-dimensional evaluation calculation on the non-dominated solution set to obtain an optimal scheduling execution scheme; and performing model feedback updating on the hybrid scheduling prediction model to obtain a target scheduling prediction model. According to the invention, multi-target balance of task completion time and energy consumption is realized, and the real-time response capability and operation stability of the system are improved.

The invention relates to the technical field of artificial intelligence and health management, in particular to a health management method and system based on multi-dimensional data analysis, and the method comprises the steps: collecting the physiological, behavior, environment and psychological data of a user, and carrying out the modeling through a recurrent neural network, and generating health state data; extracting a key causal relationship of the health risk based on a causal analysis model of a Bayesian network; comprehensively analyzing health changes through the short-term time sequence model and the long-term trend prediction model, and generating health risk data; a personalized health intervention scheme is designed by using a conditional generative adversarial network, and safe sharing and dynamic management of the scheme are realized through a block chain technology. According to the method, the health risk can be accurately identified, scientific intervention measures can be designed, the health management process is dynamically optimized, powerful technical support is provided for an intelligent old-age nursing scene, and the health management efficiency and effect are improved.

The invention discloses a multi-dimensional time sequence photovoltaic grid-connected harmonic current prediction control method, and particularly relates to the technical field of power system control. Grid-connected original multi-dimensional electrical signals are collected and decomposed into multi-dimensional frequency signals through Fourier transform, the characteristic components of an independent harmonic source are separated through the blind source separation and non-negative matrix factorization technology, and the frequency and amplitude characteristics of the characteristic components are extracted; then, a multi-dimensional time sequence prediction model is constructed through a recurrent neural network, dynamic behaviors of multiple harmonic sources are predicted at the same time, and frequency domain data are reconstructed and frequency aliasing errors are corrected in combination with high-frequency compensation and multi-scale analysis; and finally, inputting a prediction result into a photovoltaic inverter control system, dynamically generating a compensation current opposite to a target harmonic wave, realizing effective offset of the harmonic wave, accurately processing the multi-source harmonic wave problem in a complex power grid environment, and greatly improving the harmonic wave suppression capability, the electric energy quality and the system operation stability of a photovoltaic grid-connected system.

The invention provides a business vitality prediction and business district evaluation method based on multi-modal feature fusion, and relates to the technical field of spatial data analysis, and the method comprises the steps: obtaining business POI data and time sequence trajectory data of a target area, extracting the score and emotion features of user comments, and calculating the pedestrian flow distribution based on the time sequence trajectory data; then inputting the features into a recurrent neural network to predict a commercial vitality time sequence, constructing POI data into a knowledge graph, and constructing and optimizing a network reachability matrix in combination with position information and human traffic; and finally, dividing a target region by using a Louvain algorithm, fusing multi-dimensional features by edge computing nodes of each sub-region, dynamically adjusting the weight of an evaluation index through reinforcement learning, and outputting a commercial service evaluation result. By fusing multi-source data and a deep learning method, the life circle business service quality can be evaluated more accurately and comprehensively, and decision support is provided for urban planning and business development.

The invention relates to the technical field of production debugging control, and discloses a production debugging control method and system for a plastic container. The method comprises the following steps: arranging a plurality of different sensors in plastic container blow molding equipment, and simultaneously collecting blow molding process parameter data; performing wavelet threshold denoising and anomaly identification on the blow molding process parameter data to obtain a process parameter anomaly identification result; establishing a transfer function matrix between the process parameters and quality indexes according to the process parameter anomaly identification result, and obtaining a target process parameter set through multi-target optimization; and inputting the target process parameter set into a double-integral enhanced recurrent neural network for parameter regulation and control calculation to obtain a process parameter adjustment amount. According to the method, early detection and accurate positioning of process abnormity in the blow molding process are realized, the quality fluctuation risk and the defective product rate are greatly reduced, and the technical problem that different response characteristic parameters are difficult to coordinate is solved.

The invention provides a switch cabinet solid insulating material performance detection method and system based on temperature detection, and relates to the technical field of temperature detection, and the method comprises the steps: constructing a grid detection matrix on the surface of a solid insulating material, collecting real-time temperature data through a passive radio frequency identification temperature sensor, and building an initial state model; adopting a deep learning algorithm to calculate a temperature gradient value and thermal stress distribution, and constructing a performance degradation index; and the recurrent neural network is combined to predict a performance evolution trajectory, and dynamic early warning and diagnosis are realized. The performance degradation state of the solid insulating material can be accurately evaluated, the detection precision and the prediction accuracy are improved, and intelligent monitoring and early warning are achieved.

The invention discloses a fatigue driving monitoring and early warning method based on multi-dimensional feature fusion, and relates to the field of intelligent traffic and safe driving, and the method comprises the steps: carrying out the feature extraction of a collected face image of a vehicle driver through a feature extraction model based on a convolutional neural network, and obtaining a face key point; then calculating the eye closing rate, the mouth opening rate, the head posture data and the facial muscle data to finally obtain multi-dimensional features, and fusing the multi-dimensional features by using a multi-layer perceptron to obtain fused facial features; and then, a fatigue state detection model obtained based on recurrent neural network training is utilized to detect the facial feature time sequence at a plurality of continuous sampling moments so as to detect whether a fatigue driving state occurs and trigger alarm feedback. According to the method, the fatigue state of the driver can be identified and judged more accurately through combination of multi-dimensional data fusion and a time sequence analysis algorithm, and the monitoring and early warning accuracy and reliability are improved. And the system can keep high robustness even in an illumination change or complex driving environment.

The invention discloses a power efficiency optimization method based on intelligent load management. The method comprises the following steps: S1, generating an initial load characteristic matrix; s2, decomposing the initial load characteristic matrix into a short-term behavior characteristic matrix and a long-term trend characteristic matrix through pseudo-symmetry decoupling; s3, multi-level load demand prediction is carried out through a fusion structure of the multi-scale convolution and the recurrent neural network; s4, coding a cognitive gene sequence, triggering gene mutation and recombination operation when significant change occurs, and generating a load distribution strategy; s5, a load distribution strategy is transmitted into a causal reasoning and self-supervision feedback control module, and if the deviation between feedback data and a prediction result exceeds a threshold value, a self-learning mechanism is triggered; s6, dynamically distributing the load priority, and generating a decision path of load management; and S7, performing adaptive optimization through a multi-layer feature feedback closed-loop optimization system. The intelligent load management method is adopted, and the power efficiency is optimized through multi-layer feature decoupling and closed loop.

The invention discloses a power system false data injection attack detection method and system. Measurement data of a power distribution system are acquired; constructing an attack sequence, and injecting the constructed attack sequence into the obtained measurement data to simulate a false data attack; constructing a false data injection attack detection model based on the expansion causal convolution and the recurrent neural network to analyze the state change of the power distribution system; and judging an attack behavior according to the analyzed state of the power distribution system. According to the method, time sequence data can be processed, the long-term dependency relationship and the complex mode can be captured, and the response capability to the complex attack mode can be improved, so that the false data injection attack can be effectively detected and prevented, and the safety and reliability of the power distribution system can be maintained.

The invention relates to the technical field of data processing, and discloses an intelligent risk early warning method, device and equipment for a power distribution network, and a medium. Historical risk monitoring data of the power distribution network under multiple dimensions are fused through a graph convolutional network and a variational auto-encoder; a variational recurrent neural network and a long-short term memory network are trained in combination with historical risk fault data of the power distribution network, and an attention mechanism is introduced in the training process to generate a risk assessment model; acquiring real-time risk monitoring data of the power distribution network under multiple dimensions to extract multi-dimensional real-time fusion features and inputting the multi-dimensional real-time fusion features into the risk assessment model for processing to obtain a real-time risk assessment level so as to further process the multi-dimensional real-time fusion features through a multi-criterion decision analysis method and a fuzzy inference system; the target risk assessment level is obtained, the level is compared with the risk early warning threshold value, if the level exceeds the threshold value, early warning is triggered, and the accuracy of power distribution network risk assessment is effectively improved.

The invention relates to the field of energy storage frequency modulation prediction, and discloses an energy storage frequency modulation instruction prediction method and system based on a two-stage method, and the method comprises the steps: obtaining an energy storage frequency modulation instruction, and carrying out the variational mode decomposition; judging mutual attraction or mutual repulsion of the subsequences according to a variation mode decomposition result; if it is judged that the two subsequences attract each other, sequence fusion operation is carried out, and if it is judged that the two subsequences repel each other, sequence generation operation is carried out; obtaining a new sub-sequence set through sequence fusion operation and sequence generation operation; and respectively inputting the sequences in the new sub-sequence set into the recurrent neural network to obtain a final prediction result of the energy storage frequency modulation instruction. By optimizing the frequency modulation instruction in advance, the energy storage system can respond to the frequency change requirement more quickly and accurately, the frequency modulation efficiency and economic benefits of the power system are remarkably improved, and meanwhile the stability of the system and the operation efficiency of the power plant are enhanced.

The invention discloses a multi-modal content understanding method and system based on a knowledge graph, and belongs to the technical field of multi-modal content understanding, the method comprises the steps of obtaining multi-modal input data and conducting feature decoupling, semantic information in the multi-modal data and noise and redundant information peculiar to modals can be effectively separated through the feature decoupling technology, and the multi-modal content understanding efficiency is improved. The method comprises the following steps of: establishing a space-time perception graph attention network, improving the purity and semantic expression capability of features, bridging semantic gaps among different modals through the space-time perception graph attention network, realizing cross-modal semantic alignment, enhancing the generalization capability of a model, deeply mining space-time modes, relationships and anomalies in data through space-time association reasoning, and improving the accuracy of the data. The method provides support for multi-modal data analysis in a complex scene, and combines a graph neural network and a recurrent neural network to capture semantic association and spatial-temporal dynamics and optimize the performance of an inference model.

The invention provides a third-generation semiconductor device multi-physical field coupling effect dynamic monitoring method and system, and relates to the technical field of semiconductors, and the method comprises the steps: collecting the data of a temperature field, an electric field, a magnetic field and a stress field through a complementary sensor array, constructing a feature tensor, inputting the feature tensor into a pre-trained graph neural network, and extracting a coupling effect dynamic evolution rule; on-line analysis and prediction are carried out by using an adaptive time window and a recurrent neural network, a multi-field coupling correlation degree evaluation model is established, an early warning level is determined according to the deviation between the coupling correlation degree and an early warning threshold value, and device risk evaluation and maintenance suggestions are generated, so that early warning of the failure risk of the semiconductor device is realized.

The invention relates to the technical field of the Internet of Things and artificial intelligence, and discloses an equipment maintenance fault intelligent analysis method and system based on the Internet of Things, and the method comprises the steps: representing an entity and a relation as a time function, and constructing a time sequence knowledge graph; monitoring vector change in the knowledge embedding space, detecting a concept drift phenomenon and mining an equipment performance evolution mode; constructing an equipment state transition model by using a recurrent neural network, predicting the future state of the equipment and evaluating the fault risk; collecting feedback processing knowledge conflicts and updating the model; generating multi-time-scale fault risk early warning and maintenance suggestions; according to the method, the knowledge acquisition efficiency is improved, dynamic knowledge support of the full life cycle of the equipment is provided, early recognition of gradient faults is realized, the method has adaptive knowledge updating capability, multi-dimensional fault analysis is provided, maintenance resource configuration is optimized, and the equipment management efficiency is improved.

The invention provides a neuromorphic visual target tracking method and system based on image processing, and relates to the technical field of neuromorphic calculation, and the method comprises the steps: fusing the input of an event camera and a conventional image sensor, constructing a combined input tensor, and introducing a multi-scale convolution and synaptic event driving mechanism. Quick response and stable feature extraction of a high-speed moving target are realized, and coding block mistaken deletion and tracking loss caused by quick movement of the target are effectively avoided, so that system delay is reduced. Meanwhile, in combination with significance entropy difference evaluation and a dynamic brightness enhancement mechanism, the target judgment accuracy under low illumination and complex backgrounds is improved; redundant noise blocks are screened out through a significance weight function and confidence calculation, the redundancy calculation burden is relieved, and the lightweight characteristic of the system is guaranteed. A memory trajectory tensor and dynamic template adjustment mechanism based on a recurrent neural network is further introduced, time sequence consistency verification and self-adaptive updating are achieved, and the stability and robustness of the tracking process are enhanced.

The invention relates to the technical field of agricultural information, and discloses a visual analysis method and system based on an agricultural big data model. According to the system, crop growth time sequence data are processed by using a multi-scale feature extraction algorithm combining a recurrent neural network and wavelet transform, and key conversion points of a crop growth stage are automatically identified by using a deep learning architecture combining a variational auto-encoder and a hidden Markov model. A standard growth track model and a multi-dimensional deviation measurement index are constructed to evaluate the abnormal growth condition of crops, a dynamic threshold model based on reinforcement learning is applied to realize early warning of abnormal growth, and a multi-dimensional visual interface based on a clock metaphor is designed to visually display crop growth information. The crop growth stage recognition precision and early warning timeliness are improved, and the agricultural production management efficiency is remarkably enhanced.

The invention discloses a concrete internal defect quantitative detection method and system based on reflected waves, and belongs to the technical field of nondestructive testing. A reflected wave data matrix is obtained through multi-angle excitation and synchronous receiving; calculating energy characteristics of each channel, and constructing an energy response residual field; extracting waveform offset, spectrum jitter and phase change caused by defects by adopting a disturbance comparison algorithm to form a disturbance feature vector set; a defect-response function curved surface is further constructed, a defect topological structure is inversed based on gradient and curvature analysis, and defect geometric parameters are output; and finally, inputting the multi-moment defect parameters into the recurrent neural network, and predicting a defect evolution path and a failure risk. The method has high resolution and trend prediction capability, and is suitable for detection and early warning of concrete structures in bridges, tunnels and nuclear power projects.

An artificial intelligence-driven sentiment analysis system for social media monitoring, comprising: a data collection module configured to retrieve user-generated content from at least one social media platform via application programming interfaces (APIs); a preprocessing module configured to process and clean the collected data by removing irrelevant content, tokenizing the text, removing stop words, and assigning sentiment indicators to emojis, slang, and abbreviations; a sentiment analysis engine comprising one or more deep learning models, wherein the deep learning models include recurrent neural networks (RNNs), long-short-term memory networks (LSTM), and transformer-based models configured to analyze the preprocessed data and classify the content into one or more sentiment categories; an emotion recognition module configured to detect emotional tones within the content and classify them into predefined emotional categories; a real-time data processing unit configured to continuously monitor social media content and provide real-time insights into moods and emotions; a feedback loop module configured to update and refine the deep learning models through reinforcement learning based on new data inputs and sentiment classification results; and a user interface module configured to visualize sentiment and emotion analysis results in real time, including heatmaps, sentiment trend analysis, and emotional breakdowns.

The invention provides a hydro meteorology early warning method for an offshore oil and gas platform, and belongs to the technical field of offshore hydro meteorology. Extreme weather events are identified by adopting minimum probability abnormal event identification vectors to match abnormal characteristic parameters, and abnormal signal characteristic parameters are input into an ocean dynamics prediction model to calculate real-time sea condition parameters; calling a multi-temporal-spatial-scale early warning fusion matrix to combine with a wavelet decomposition technology and a recurrent neural network to realize multi-scale information integration, analyzing an environmental parameter change trend through a sea condition jump identification model and triggering an emergency response, dynamically adjusting system parameters according to a stability evaluation index vector, and optimizing prediction precision by adopting an early warning residual value compensation matrix. And finally, multi-level early warning information is generated and a real-time early warning notification is sent to an operator, so that the technical problem of insufficient early warning precision of an offshore oil and gas platform hydro meteorology early warning system in multi-spatio-temporal scale data fusion processing is solved.

The invention discloses an AI search recommendation method, system and device combined with commodity semantic understanding and a medium, and belongs to the technical field of commodity recommendation. The method comprises the steps that semantic vectors of commodity titles and descriptions are generated; generating a feature vector of the commodity image; fusing the semantic vector of the commodity title and description with the feature vector of the commodity image to generate a comprehensive semantic vector of the commodity; generating a user interest vector; updating the user interest vector by using a recurrent neural network based on the user interest vector; and carrying out joint modeling on the comprehensive semantic vector of the commodity and the updated user interest vector to generate a recommendation result. According to the method, a content semantic driving modeling mode is adopted, recommendation judgment can still be conducted through the deep matching relation between commodity image-text semantics and user behavior preferences even under the condition that user historical behaviors are limited or new commodities are online, good cold start adaptability is achieved, and the method is suitable for popularization and application. And meanwhile, high recommendation difference and content diversity are shown for different user groups.

The invention discloses a high-precision AI positioning method and system based on spatial multi-modal data fusion, and belongs to the technical field of fusion positioning, and the method specifically comprises the steps: collecting satellite positioning signals, inertial measurement data, visual images, laser radar point clouds, high-precision maps and other spatial multi-modal data; analyzing visual image illumination distribution, laser point cloud atmospheric particle density distribution and a satellite signal multipath reflection path to generate an environment physical attribute parameter set; inputting the original data of each modal into a correction function bound with environmental parameters, and outputting a standardized feature vector subjected to error compensation; and finally, inputting the recurrent neural network with space-time memory, and outputting the six-degree-of-freedom pose of the carrier by dynamically adjusting each modal feature fusion weight coefficient. According to the invention, through deep fusion of multi-modal data and environmental physical attributes, the problems of poor environmental adaptability and insufficient dynamic adjustment in the prior art are solved, and the precision and robustness of positioning in a complex scene are improved.

The present invention relates to urban rainfall runoff pollution prediction in urban water systems, and provides an urban rainfall runoff pollution prediction method based on integrated rolling decomposition method and deep learning algorithm. A rolling decomposition method is firstly used to decompose rainfall runoff sequence data into different sub-sequences; then decomposition is sequentially performed on added data, and future data is excluded, to prevent information leakage; a recurrent neural network is used to model and predict the sub-sequences; and finally, predicted results of the sub-sequences are summed to obtain the predicted result of rainfall runoff pollution.

The embodiment of the invention provides a barrel service life prediction method and device based on gradient enhancement and a quantile recursive network, and the method comprises the steps: collecting the multi-source sensor data of the whole life cycle of a barrel, and carrying out the preprocessing, and forming a standardized data set; constructing an XGBoost model, performing feature importance analysis on the standardized data set through the XGBoost model, screening key features based on an analysis result to obtain a data set after feature selection, and performing optimization training on the XGBoost model through gradient lifting and regularization methods; dividing the data set after feature selection into a training set and a test set; in combination with a quantile regression method, constructing a life prediction model based on a quantile recurrent neural network, training the life prediction model by using the training set, and training and optimizing model parameters through a bifurcated sequence to obtain an optimized barrel life prediction model; and inputting a test set into the model, outputting a residual service life prediction result containing a prediction value and a confidence interval, and evaluating the prediction performance.

The invention relates to the technical field of constructional engineering, and discloses a bridge hanging basket installation accurate positioning automatic operation method, which comprises the following steps of: acquiring image data of the position, the speed, the posture and the construction environment of a bridge hanging basket in real time by installing a high-precision sensor and a visual system; analyzing the acquired image data by using a deep learning model, and identifying and estimating the spatial position and angle of the hanging basket; the system comprises a high-precision sensor and an image acquisition system which are used for acquiring the position, attitude and environmental data of the hanging basket in real time; the deep learning module comprises a convolutional neural network and a recurrent neural network and is used for recognizing the position of the hanging basket in real time and predicting future deviation; and the nonlinear optimization module is used for calculating optimal control input in real time according to the optimization objective function and correcting the optimal control input. By adopting the technical scheme of the high-precision sensor and the image acquisition system, real-time accurate monitoring of the position, the posture and the construction environment of the hanging basket is achieved.

The invention relates to a multi-source disturbance perception and self-adaptive regulation and control technology in an electroplating production process, and aims to solve the problem that the process uniformity is difficult to guarantee due to diverse and dynamic changes of disturbance sources in an electroplating process. The invention provides a regulation and control method taking multi-dimensional disturbance sensing signal acquisition, preprocessing, recurrent neural network prediction, spatial distribution modeling and meta-learning reinforcement learning regulation and control as the core. The method comprises the following steps: acquiring various disturbances and auxiliary parameters such as fluid, temperature and current in real time, performing denoising, normalization and time sequence alignment processing, dynamically predicting a disturbance trend by using a recurrent neural network, and combining with process parameters to form high-expression feature input. And training the reinforcement learning regulation and control model by using a meta-learning algorithm to realize rapid adaptive optimization of the process parameters in the new disturbance scene. According to the system, through real-time feedback and dynamic self-optimization, the electroplating uniformity and the production stability are improved, and the robustness and the self-learning ability to a complex disturbance environment are effectively enhanced.

The invention provides a robot social adaptive navigation knowledge learning and migration method and system, and relates to the field of mobile robot navigation. Aiming at the problems that an existing path planner lacks time sequence memory and neglects pedestrian social intent, a man-machine co-fusion scene is constructed, a training set containing an expert demonstration path is made, and a recursive generation model is input; designing a recurrent neural network embedded RRT, generating an RNN-RRT planner, and fusing historical information and pedestrian convergence probability in training; new scene loading training parameters are finely adjusted to realize knowledge migration, loss convergence or output RNN final parameters after reaching a preset round number. According to the method, the path anthropomorphism and generalization ability are improved, and the method is suitable for complex human-computer interaction scenes.

The invention relates to a quay crane RTK positioning precision evaluation method and system and a storage medium, and the method comprises the steps: L1, when an unmanned container truck carries out the container loading and unloading operation under a quay crane, the quay crane RTK feeds back the positioning value of a vehicle according to a fixed frequency, and obtains the data information of the positioning value of the vehicle within a fixed time; l2, based on the data information of the positioning value of the vehicle within the fixed time, predicting the positioning value of the vehicle by adopting an improved ELMAN dynamic recurrent neural network prediction algorithm optimized based on a particle swarm optimization algorithm to obtain the data information of the predicted positioning value of the vehicle; and L3, based on the predicted data information of the positioning value of the vehicle, optimizing the positioning value of the vehicle by adopting a zebra optimization algorithm of a self-adaptive dynamic learning factor based on global search to obtain the optimized data information of the positioning value of the vehicle. The alignment success rate of the unmanned container truck can be increased, and harbor area operation can be prevented from being affected.

The invention relates to the technical field of energy storage frequency modulation control, and discloses an energy storage frequency modulation instruction control method and system, and the method comprises the steps: obtaining an energy storage frequency modulation instruction, and carrying out the variational mode decomposition; calculating a balance value of the sequence after variational mode decomposition, and generating a supplementary sequence according to the balance value; processing the supplementary sequence to obtain a de-filling sequence, correspondingly adding the sequence after variational mode decomposition and the supplementary sequence, and then forming a new sub-sequence set with the de-filling sequence; and respectively inputting the sequences in the new sub-sequence set into the recurrent neural network to obtain a final control result of the energy storage frequency modulation instruction. According to the method, the problems of high prediction difficulty and low precision caused by too strong fluctuation of the frequency modulation instruction sequence in the prior art are solved, the irregularity and fluctuation of the original sequence are remarkably reduced, the result is more accurate and reliable, the frequency modulation requirement can be responded in advance, and the immediate response capability of the system is improved.

The invention relates to the technical field of computer vision and multi-sensor data fusion, in particular to a multi-sensor image fusion obstacle real-time detection and tracking system, which comprises the following steps of: firstly, fusing data of a camera, a millimeter wave radar and a laser radar, and extracting and fusing multi-modal features; performing multi-target tracking based on a recurrent neural network: generating a target state through space-time modeling, associating a target with a historical track by using an attention mechanism, and maintaining track consistency; and the system performs semantic classification and interaction analysis on the obstacle, predicts the movement track of the obstacle, realizes deep semantic understanding, and finally outputs the identity label and the complete historical track of the obstacle.