97results about How to "Achieve deep integration" patented technology

The invention relates to an underwater acoustic target recognition system and method based on multi-modal depth feature fusion, and belongs to the field of underwater acoustic target recognition. The method comprises the following steps: firstly, performing preprocessing on an obtained underwater acoustic target original audio and associated metadata, and constructing a multi-modal data set; and then the constructed multi-modal sample is input into an identification model, the model extracts deep representation of each modal through a multi-branch feature coding network, depth alignment and complementary aggregation of different modal features are realized by using a cross-modal cross attention mechanism guided by potential query, and a target identification result is output based on a decision network of mixed experts. According to the method, experimental verification is carried out on two disclosed underwater acoustic data sets, the experimental result verifies the effectiveness of the multi-modal deep fusion and hybrid expert adaptive decision strategy adopted by the method, and through mining the complementary advantages of acoustic features and semantic priori, the multi-modal deep fusion and hybrid expert adaptive decision strategy is obtained. And the robustness and generalization ability of the underwater acoustic target recognition system in the strong-noise and multi-working-condition environment are remarkably improved.

The invention relates to a mapping system and method for flight inspection of indoor equipment, the system comprises a data acquisition end and an offline data processing end, the data acquisition end comprises an unmanned aerial vehicle and a sensing module, and the sensing module comprises a laser radar module, a binocular camera module, an inertial measurement module and an airborne recording module; the method comprises the following steps: synchronously acquiring multi-source data through layered flight; performing space-time alignment and preprocessing; self-adaptive sampling is carried out based on point cloud geometric prior guide image features, and laser-vision joint features are generated; fusing point cloud registration, joint features and inertial data, and carrying out joint estimation on adjacent frame pose increments; performing loopback detection based on the key frame to generate a closed-loop constraint; constructing and optimizing a pose map to eliminate cumulative drift; and finally fusing to generate a global point cloud map. According to the method, the space consistency and geometric accuracy of the map can be effectively improved, and a reliable data basis is provided for digital modeling and intelligent operation and maintenance of indoor equipment inspection.

The invention discloses a combined preprocessing and fusion decoding method based on electroencephalogram signals and functional near-infrared signals, which comprises the following steps of: synchronously acquiring the electroencephalogram signals and the functional near-infrared signals, taking a task trigger event as a unified time reference, and carrying out basic preprocessing of time alignment and modal self-adaption on two modal signals to obtain a functional near-infrared signal; a pre-trained cross-modal noise joint modeling module is utilized to perform joint modeling on cross-modal joint noise features caused by a common noise source, collaborative denoising processing is performed on multi-modal signals based on the cross-modal joint noise features, multi-modal denoising representation is obtained, a cross-modal feature interactive modeling mode based on an attention mechanism is obtained, and the multi-modal noise is obtained. And dynamically modeling the correlation between different modal features, realizing deep fusion of multi-modal complementary information, carrying out decoding processing based on the fused features, and outputting a corresponding brain-computer interface control instruction or task identification result. According to the method, the decoding accuracy and robustness of the brain-computer interface system in a complex task scene are enhanced.

The invention relates to the technical field of digital archive management and information retrieval, and discloses a digital archive multi-modal data semantic enhancement fusion retrieval method and system.The method comprises the steps that a policy cycle time axis and a policy term evolution graph are constructed, tense logical reasoning is conducted on archive seals, and permission effectiveness evolution is derived; the temporal permission feature vector and the content semantic vector are fused to generate a multi-modal representation vector, and cross-policy-cycle semantic enhancement retrieval is realized by combining query expansion and temporal permission filtering, so that the problems of missing detection and misjudgment of policy and regulation archives in seal permission historical evolution and term cross-cycle retrieval are solved.

The invention relates to the technical field of ablation path planning, in particular to a liver cancer thermal ablation path planning system based on deep reinforcement learning. The system comprises a target and dangerous area mask generation module, a pipeline distance field construction module, a thermal field prediction module, a breathing deformation prediction module, a multi-scale feature fusion module, a behavior preference reward module, a composite reward function module and a decision output module. Clinical pain points such as breathing interference, dangerous area avoidance and individualized adaptation are broken, the precision and safety of the liver cancer thermal ablation operation are improved, normal tissue damage and complications are reduced, intelligent decision support is provided for interventional therapy, and liver cancer minimally invasive therapy is promoted to be upgraded to precision and intelligence.

The invention relates to the technical field of industrial big data, in particular to an Internet of Things sensing and data monitoring integrated system, which comprises an Internet of Things sensing module used for acquiring a sensing parameter set of a distributed simulation unit of an intelligent power plant and a historical performance evolution trend reflecting a dynamic evolution rule; the twin mapping module is used for realizing signal space-time synchronization by utilizing a time service protocol, calling a multi-physics field evolution model to calculate thermal stress offset so as to execute numerical compensation, and generating a high-fidelity physical feature vector; the collaborative decision-making module shares observation information by using a multi-agent distributed collaborative architecture, identifies an electromagnetic-dynamic coupling response mode through a joint reward function and outputs a collaborative optimization strategy; and the execution module is used for generating an evaluation result by calculating the numerical deviation between the physical feature vector and the design envelope, and adjusting the excitation control variable according to a strategy until the system converges to a target equilibrium value. According to the method, the steady state monitoring of the power system is realized through twin mapping and collaborative decision.

The invention relates to an intelligent multi-source navigation method based on inertia and Beidou / visual information fusion, which comprises the following steps: firstly, establishing a multi-source information system of an inertial navigation system, and embedding an IMU (Inertial Measurement Unit) multi-parameter online calibration mechanism based on carrier dynamics in inertial navigation solution; establishing a tight coupling fusion framework taking an INS resolving result as a reference, and constructing tight combination measurement with the original observed quantity of the BDS and constructing re-projection error measurement with the visual feature points by utilizing INS high-frequency pose information; an intelligent adaptive filter based on inertial dynamics error model constraint is adopted, an INS error equation is used as a state prediction model, a Sage-Husa algorithm is introduced to estimate system noise statistical characteristics in real time, estimated error parameters are fed back to INS calculation through closed-loop correction, real-time correction of navigation results is achieved, and the system noise statistical characteristics are estimated in real time. Through deep tight coupling fusion and intelligent adaptive processing, navigation precision and system reliability in a complex environment are effectively improved.

The application relates to the technical field of hydrological prediction and deep learning, and provides a conceptual hydrological neural network hydrological prediction method and system considering space-time dynamic parameters, the method comprising the following steps: acquiring digital elevation model data of a target region, extracting a river network structure, and dividing the target region into a plurality of sub-basin units; constructing a sub-basin feature set and a river channel feature set based on underlying surface information data of the target region; constructing an original sample set based on historical hydrological and meteorological data of the target region; fusing the sub-basin feature set and the original sample set to obtain a standardized time series sample set; training a conceptual hydrological neural network model based on the standardized time series sample set, the original sample set, the sub-basin feature set and the river channel feature set to obtain a trained conceptual hydrological neural network model; and inputting real-time hydrological and meteorological data into the trained conceptual hydrological neural network model to obtain hydrological prediction information of the target region. Thus, the hydrological prediction accuracy is effectively improved.

The invention discloses a method for constructing, updating and retrieving an action memory library, which belongs to the technical field of computer vision and comprises the following steps of: constructing a training data source with time sequence diversity; initializing an action memory library containing a plurality of learnable prototype matrixes and a double-flow interaction network; memory bank evolution is executed, an action prototype is retrieved by utilizing a query stream, a current memory state is dynamically generated by combining a memory state updating gate mechanism with a historical state, and dynamic memory is injected into a feature space by utilizing memory driving graph convolution; synchronously updating parameters based on multi-target loss, and driving a memory bank to evolve into optimal structured prior; and finally, performing structured reasoning on the to-be-detected sequence by using the optimal memory bank. According to the method, structured priori is constructed by mining a spatio-temporal topology mode of a human body action sequence, and hierarchical memory evolution and double-flow depth interaction are combined, so that the problem of depth ambiguity in a monocular vision task is effectively solved, geometric structure distortion is corrected, and the accuracy of action posture estimation is remarkably improved.

This invention discloses a navigation method and system for offshore wind power booster stations based on multimodal information interaction, belonging to the field of artificial intelligence and robot navigation technology. It achieves visual-language modality alignment through the CLIP model, constructing an end-to-end navigation parameter generation model. This model can adaptively generate AGV motion parameters based on natural language commands and scene images, enabling efficient and accurate inspection of offshore wind power booster stations. By establishing a multimodal dataset and associating language commands with AGV motion parameters, this invention solves the error accumulation problem caused by multi-stage processing in traditional methods, improving navigation accuracy and adaptability. Furthermore, this invention employs a multi-task loss function optimization model, including contrastive loss and motion parameter regression loss, enabling the AGV to adaptively adjust motion parameters according to environmental changes, thus improving the efficiency and safety of offshore wind power booster station inspections.

The invention provides a municipal road maintenance demand prediction method and system based on machine learning, and relates to the technical field of machine learning, and the method comprises the steps: obtaining the service starting time of a municipal road and the passing frequency and full load proportion of heavy-load vehicles in a plurality of sampling periods; difference calculation is carried out on the service starting time and the sampling time points in each sampling period, and a space-time incidence matrix is constructed; taking an attenuation coefficient in the space-time incidence matrix as a dynamic regulation factor to obtain a bit stream sequence; mapping the bit stream sequence as an observation sequence to a preset road grid map, and determining an optimal transfer path through optimal path planning; and determining a grid step length for transferring the road state to the maintenance state at the current moment from the optimal transfer path, and determining maintenance demand prediction information comprising a maintenance priority and a predicted maintenance date. The precision, objectivity and foresight of municipal road maintenance demand prediction are realized.

The invention relates to a GIS equipment state monitoring method and device, computer equipment and a medium. The method comprises the following steps: acquiring multi-source equipment data of target GIS equipment; performing feature extraction on the multi-source equipment data to obtain equipment feature data; obtaining internal state data of the equipment according to the equipment characteristic data and a pre-constructed equipment data twin model; the equipment data twin model is constructed based on equipment attribute data of the target GIS equipment; obtaining a state monitoring result of the target GIS equipment according to the equipment characteristic data, the equipment internal state data and a state monitoring model; and the state monitoring result is used for representing the health degree of the target GIS equipment and the possibility of various preset equipment faults, so that the accuracy of the state monitoring result of the target GIS equipment is improved.

The invention provides a board card defect detection method and system based on image processing, and the method comprises the steps: collecting an original image of a to-be-detected board card, and sequentially carrying out the image preprocessing and image detail splitting of the original image, so as to obtain a background image and a detail image; performing first enhancement processing on the background image to obtain an enhanced background image, and performing second enhancement processing on the detail image to obtain an enhanced detail image; fusing the enhanced background image and the enhanced detail image to obtain a target enhanced image; performing iterative enhancement on the depth image based on the original image to obtain an enhanced depth image; the enhanced depth image and the target enhanced image are fused to obtain a final fused image, the final fused image is input into a defect detection model for defect detection, and a board card defect detection result is output, and the problems that in the prior art, detection precision is low, robustness is poor, and three-dimensional defects cannot be detected are effectively solved.

The invention discloses a refined behavior recognition method and system for a virtual teaching scene, and solves the technical problem that the precision of a final behavior recognition result is poor due to an existing refined behavior recognition method for the virtual teaching scene. The method comprises the steps that a virtual teaching scene behavior image is acquired, the virtual teaching scene behavior image is input into a behavior measurement model based on multilayer perception self-adaption, and the behavior measurement model based on multilayer perception self-adaption comprises a feature extraction network, a feature enhancement network and a posture decoding network; performing multi-scale feature extraction on the virtual teaching scene behavior image through a feature extraction network, and outputting multi-scale fusion features; performing feature enhancement on the multi-scale fusion features by adopting a feature enhancement network, and outputting target enhancement features; and inputting the target enhanced feature into a posture decoding network for posture decoding, and generating a behavior recognition result.

The invention discloses a blue army task planning method based on knowledge and data driving, and relates to the technical field of military simulation modeling and computer simulation, and the blue army task planning method comprises a sensing layer, a fusion layer, a reasoning layer, a planning layer, an optimization layer and an output layer. Bidirectional interaction between the layers is achieved through data interfaces, and deep fusion of knowledge and data in the whole process is ensured; by constructing a dynamically updated knowledge graph and a knowledge-data mapping mechanism, domain knowledge can guide data modeling in the whole process, meanwhile, the data can optimize a knowledge system in real time, the limitation of a single driving mode is solved, and through testing, in a complex military drill scene, the knowledge and data can be fully fused. The matching degree of the fusion-driven planning scheme and the actual situation reaches 92%, is improved by 25% compared with a knowledge-driven scheme, and is improved by 30% compared with a data-driven scheme.

The application discloses infrared FPA-on-MEMS technology, belongs to the field of optical imaging and micro-electro-mechanical system integration technology, is used for manufacturing of infrared FPA-on-MEMS chip, utilizes the design of electrothermal-electrostatic dual driving module manufacturing process and the integration of infrared FPA unit on the middle platform of electrothermal-electrostatic dual driving module, can realize the coupling design of infrared FPA unit and MEMS driving technology on the chip level, and obtains infrared FPA-on-MEMS chip with automatic displacement control capability.The infrared FPA-on-MEMS technology of the application can effectively overcome the problems of large volume, slow speed, high power consumption and complex system caused by the fact that the traditional infrared imaging system relies on external optical mechanical components to realize functions such as anti-shake, scanning and zooming, improves the integration and reliability of the infrared thermal imaging system, breaks through the application bottleneck of the infrared thermal imaging technology, greatly improves the system integration and comprehensive use performance of the infrared thermal imaging system, and has excellent application prospect.

The invention discloses an end-to-end automatic driving track determination method and a related product. In the scheme, based on sensor data, map information and a navigation target, an environment context feature vector of the autonomous vehicle is obtained; performing automatic driving track flow direction prediction based on the current track state information, the current time and the environment context feature vector of the automatic driving vehicle, and determining an initial flow direction prediction result; correcting the initial flow direction prediction result by using the constraint correction flow direction to obtain a target flow direction prediction result; the constraint correction flow direction is determined based on the physical constraint and the security rule constraint; the physical constraint is constructed based on a Riemannian metric inverse matrix; the security rule constraint is constructed based on the semantic potential field gradient; and processing the target flow direction prediction result to obtain a trajectory determination result of the autonomous vehicle. Compared with the problems of low planning precision and poor real-time response capability in the prior art, the method has obvious advantages.

The invention discloses a malignant tumor image fusion analysis method and system based on graph structure consensus synergy, and the method comprises the steps: receiving multi-modal medical image data, extracting an initial high-dimensional feature vector, and projecting the initial high-dimensional feature vector to a plurality of independent potential semantic subspaces; a feature influence value and a dimension discrimination score are calculated through a dual significance evaluation mechanism, node features are weighted, edge weights are modulated, and a sample specificity initial multi-path diagram is constructed; after node features are linearly expanded, an affinity matrix is generated and sparsified, and a robust optimization graph is output through graph structure enhancement and contrast learning; coding each plane feature by adopting a graph attention network, and generating a fusion feature through high-discrimination feature splicing and weighted summation of other features; and constructing a global aggregation graph by taking the fusion features as nodes, and aggregating global information to output a diagnosis result. The system correspondingly comprises a multi-modal feature extraction unit, a multi-modal feature analysis unit and the like, multi-modal image semantic synergy and structure optimization are achieved, and malignant tumor diagnosis accuracy is improved.

This application discloses a computer network intrusion detection and defense method based on anomaly data analysis, relating to the field of computer network technology. The method includes: collecting a multi-dimensional dataset; extracting the correlations between anomaly data to form an anomaly correlation rule set; fusing multi-source features to form an anomaly feature set; establishing an intelligent agent model; the intelligent agent executing an attack task; inputting the anomaly data set into the intrusion detection model to obtain the first-stage detection result; comparing and analyzing the first-stage detection result with the first-stage simulated attack result to obtain feedback information; optimizing the second-stage attack strategy and executing the optimized second-stage attack strategy to obtain the second-stage detection result; extracting anomaly features to form a final anomaly feature set; generating repair instructions based on the final anomaly feature set and executing the repair instructions to repair the computer network. This application's method overcomes the limitations of single attack simulation and deepens the two-stage attack optimization, improving detection accuracy.

The invention provides an intelligent medical insurance compliance examination method and system based on graph retrieval enhanced generation, relates to the field of medical informatization and artificial intelligence, and solves the technical problem that recessive medical insurance violation behaviors with context dependence and structure combination cannot be effectively identified in the prior art. The method comprises the following steps: constructing a compliance knowledge graph; extracting clinical entities from the diagnosis and treatment records, and mapping the clinical entities to corresponding nodes in the compliance knowledge graph; based on the mapped clinical entity as a query starting point, executing multi-hop graph retrieval in the compliance knowledge graph to obtain a policy rule sub-graph related to the current diagnosis and treatment behavior, and generating a compliance review result through structured reasoning; constructing a violation risk propagation model based on the historical auditing data, and performing violation risk assessment on the diagnosis and treatment behaviors to obtain a risk early warning level; and generating an audit priority queue for the plurality of to-be-audited diagnosis and treatment records based on the violation confidence and the risk early warning level. The method and the device are used in a medical insurance compliance examination process.

The application discloses a kind of sepsis multiple complications prediction method, system, storage medium and equipment, belong to intelligent medical technical field, method includes the following steps: S1. acquisition sepsis patient multidimensional time series physiological data;S2. medical perception feature vector is constructed to the time series physiological data;S3. sepsis multiple complications prediction model is constructed and introduced physiological logic space-time composite mask matrix;S4. the output multiple complications prediction probability using hierarchical expert hybrid network H-MoE;S5. the model is trained using dynamic weighted loss function based on task uncertainty;S6. risk output and feature contribution degree explanation;The model of the application is introduced by physiological logic mask and H-MoE architecture, realizes the accurate, synchronous prediction of sepsis complex complications, can effectively filter redundant features, and reveal the complex correlation therebetween, provide quantitative decision support for clinician, to realize early intervention and individualized treatment.

The application discloses a kind of dynamic spatio-temporal graph learning microservice exception detection methods for multi-modal data, which comprises: collecting data information from the microservice system to be detected and inputting into computer system and carrying out data preprocessing, obtaining multi-modal data;The multi-modal data is processed based on the sliding window mechanism, and a dynamic dependency graph sequence is constructed;Based on dynamic dependency graph sequence, spatial relationship modeling and time series dynamic evolution modeling are performed, a representation integrating time series information is generated, and a graph-level representation vector containing complete spatio-temporal context is generated;The graph-level representation vector is input into a deep vector data description anomaly detection model based on contrast learning, and the detection result is output.The microservice exception detection based on multi-modal dynamic spatio-temporal graph learning and contrast enhancement SVDD of the application explicitly captures the dynamic evolution characteristics of microservice topology structure, and realizes the deep fusion of spatio-temporal coupling characteristics, while suppressing the decision boundary ambiguity problem in unsupervised detection.

The application belongs to the technical field of water industry, and relates to a multi-source data semantic interaction method and system for the water industry, comprising: analyzing and processing multi-source monitoring data collected in real time to obtain a structured semantic data set containing multiple parameters; based on a preset early warning threshold, performing abnormality judgment on a target parameter; the target parameter is a specific monitoring parameter selected in the structured semantic data set; when the target parameter is abnormal, based on a pre-constructed correlation retrieval fusion model and the structured semantic data set, calculating the causal correlation strength of a first-level correlation parameter directly correlated with the abnormal target parameter; and according to the causal correlation strength of the first-level correlation parameter, the structured semantic data set and a parameter correlation rule, dynamically reasoning the abnormal reason of the abnormal target parameter. The application breaks through the semantic barrier of multi-source monitoring data, improves the accuracy of alarm, can accurately locate the abnormal reason, realizes efficient decision-making, and adapts to the multi-scene requirements of the water industry.

The invention discloses a sepsis risk multi-modal prediction method and system based on deep learning, and belongs to the technical field of intelligent analysis of medical data, and the method comprises the following steps: a multi-modal data collection step: collecting vital sign sequences, bedside instant inspection, laboratories, microbial culture, images and medical history data; in the time sequence feature coding step, a time sequence embedded vector is extracted through a multi-scale time sequence coding network; the cross-modal fusion prediction step is based on the dynamic attention mechanism fusion features and predicts the sepsis occurrence probability; the organ function dynamic evaluation step is used for estimating a missing test result and calculating an organ function score; the risk layering early warning step outputs four-level risk layering results and feeds back optimized coding parameters, early warning can be output 4-8 hours before the diagnosis standard is met, and compared with qSOFA, the AUC is improved by 0.15 or above.

This invention discloses a perspective tracking method, system, device, and storage medium for commission calculation. The method includes: responding to a target commission scenario selection operation, loading metadata of the basic dataset corresponding to the target commission scenario, and generating a configuration workbench based on the metadata; receiving a dimension setting operation based on the configuration workbench, determining the dimension information corresponding to the commission calculation according to the dimension setting operation, and configuring indicator rules using a dual configuration mode; parsing the indicator rules, and calling the basic dataset to perform commission calculation based on the dimension information and indicator rules to obtain a commission result set; responding to a perspective tracking request for the commission result set, and tracing back from the commission result set to the original data corresponding to the commission calculation through hierarchical drill-down and association graph construction to achieve perspective tracking of the commission calculation process. This method enables deep integration of flexible visualization configuration and data perspective tracking in commission calculation.

The application discloses a labor risk assessment system based on deep learning judgment, belongs to the technical field of artificial intelligence and labor risk assessment, constructs an end-to-end technical architecture of "multi-modal feature fusion-multi-task deep learning-double path fusion analysis", maps multi-source labor data to the same vector space through multi-modal feature fusion, realizes dynamic evolution of multi-dimensional risk assessment standards by combining multi-task machine learning and reinforcement update, constructs a pre-deployment and in-service double-path judgment mechanism, fuses the evaluation results of three dimensions of the deployment unit risk, the deployer personal risk and the post risk, outputs recruitment decision suggestions and company overall labor operation risk rating, not only realizes a fundamental leap of labor risk assessment from static rule statistics to dynamic intelligent decision, but also realizes a closed loop coverage of labor risk from "pre-entry" to "in-process monitoring", and provides accurate and differentiated decision support for enterprises at different stages.

The invention, which relates to the technical field of energy chemical trade monitoring, discloses a global energy chemical trade cargo flow monitoring system based on an AI and a business condition knowledge graph, comprising a multi-source data acquisition layer, a data processing and storage layer, a knowledge graph construction and reasoning layer, and an AI application service layer. According to the system, AIS data, business condition data and other multi-dimensional information are acquired through a multi-source data acquisition layer, cleaning and fusion are carried out through a data processing and storage layer, a dynamically-associated business condition knowledge graph is established through a knowledge graph construction and reasoning layer, voyage number event identification, cargo flow prediction and risk early warning are achieved through an AI application service layer, and the risk of the business condition knowledge graph is predicted. The problems of data island, single-mode analysis limitation and insufficient dynamic reasoning ability of an existing monitoring system are solved, the accuracy, the real-time performance and the commercial insight ability of cargo flow monitoring are improved, and the method can be widely applied to scenes such as global energy chemical trade supervision and enterprise supply chain optimization.

The invention relates to the technical field of data processing, and discloses an end-to-end information content classification self-improvement method, device and equipment and a medium, and the method comprises the steps: receiving end-to-end original information, and carrying out the preprocessing of group sending information recognition and marking on the original information; classifying the preprocessed original information by using a pre-trained classification model to obtain a white sample and a black sample; identifying potential novel black samples and corresponding types in the white samples based on preset black sample feature constraint conditions; when the novel black sample data after enhancement meets a preset condition, adding the novel black sample after data enhancement into a historical training data set of a classification model to form an updated training data set; and retraining the classification model based on the updated training data set so as to update the classification model. The problems that in the prior art, group sending features are not considered, and updating of existing types and increasing and updating of unknown types cannot be dynamically achieved are solved.

The application relates to the technical field of electronic commerce, and discloses a SaaS template intelligent generation and adaptation method for a payment scene. The method first acquires a target payment channel operation sequence and a cross-currency exchange rate, calculates a comprehensive channel health degree and an exchange rate fluctuation rate to construct a multi-dimensional state feature vector; secondly, a symmetric semi-positive definite component interference matrix updated offline is extracted, the feature vector is mapped into a benchmark prediction probability, and multi-dimensional feature cross coupling is combined to calculate an expected comprehensive utility value of a candidate template with a non-negative interference penalty term; subsequently, a dynamic strategy exploration temperature clamped by a threshold is calculated by fusing features, and an adaptation template is generated by sampling in combination with the utility value; and finally, actual conversion results are collected, the order amount proportion with a maximum threshold cut-off is taken as a value weighting factor, and a value weighting approximate gradient is used to perform online adaptive closed-loop updating on a conversion matrix. The application effectively quantifies component bottom layer conflicts, and realizes adaptive closed-loop optimization of a payment interface in a high dynamic environment.

The invention discloses a multi-sensor management and control method based on multi-agent deep reinforcement learning, and the method comprises the steps: firstly building a state equation and a motion model of a target, and an observation model and a motion model of a sensor node; then constructing a multi-agent strategy learning module; the multi-agent strategy learning module adopts an MAPPO algorithm; then, each intelligent agent processes measurement information in the vision field range of the intelligent agent to obtain a multi-dimensional matrix state diagram and a search utility diagram; and splicing the two information with the state information of the agents, inputting the spliced information to a multi-agent strategy learning module, outputting and executing the control action of each agent at the next moment, and repeating the steps to realize the global estimation of the target in the monitoring area. According to the method, while the bottleneck of centralized calculation is effectively avoided, the estimation capability of target information outside a visual field is expanded, and deep fusion of GM-PHD probability information and a reinforcement learning reward mechanism is realized, so that the global estimation and tracking performance of a system on multiple targets is remarkably improved.