330 results about "Adaptive decision making" patented technology

The invention relates to the technical field of integrated circuits and computer networks, and particularly discloses a transmission control and intelligent scheduling system for an integrated chip, and the system sets a dual-mode decision and dynamic switching mechanism for each routing node. Calculating dynamic characteristic parameters including an instantaneous value, a first-order trend and a second-order acceleration; when the parameter is matched with a pre-stored abnormal feature set and the load exceeds a threshold value, the node is immediately atomized and switched to a predefined security scheduling strategy loaded from a shared storage area, and otherwise, the node generates a scheduling decision according to self-adaptive decision logic continuously optimized based on historical performance feedback; all the nodes carry out data packet forwarding control according to the current execution strategy; and the system also periodically realizes federated global knowledge evolution according to the quality evaluation result of the self-adaptive decision of each node.

The invention belongs to the technical field of artificial intelligence and computer vision, particularly relates to a visual processing edge adaptive decision-making method for resource and task perception, and aims to solve the problem of scheduling mismatch caused by resource dynamic change and task demand diversity in visual task processing in an edge computing environment. The method comprises the following steps: collecting multi-dimensional resource state data of edge nodes in real time to form a resource state vector with high time resolution; analyzing the visual task request, and constructing a quantifiable task feature vector; and establishing a resource-task association mapping model based on a dynamic weight distribution mechanism. The method also supports cross-edge domain collaborative decision, and processes a pipeline dynamic reconstruction and security isolation mechanism. According to the technical scheme, the fluctuation of the resource utilization rate is reduced to 15% or below, the average task processing delay is reduced to 60%, the scheduling satisfaction degree is improved by 40% or above, and the self-adaptability and the service quality guarantee capability of the edge vision system are remarkably enhanced.

The invention discloses a customized production-oriented edge node lightweight AI model adaptive compression method, which belongs to the technical field of intelligent manufacturing and edge computing, and comprises the following steps of: dynamically integrating compression strategies such as pruning, quantification and knowledge distillation by analyzing demand constraints and edge node hardware resources of customized production tasks; constructing an adaptive decision engine by utilizing reinforcement learning and Bayesian optimization, and generating an optimal compression scheme; in the deployment stage, compression parameters are dynamically adjusted through real-time monitoring and a closed-loop feedback mechanism, and the balance of model precision, reasoning efficiency and resource occupation is achieved. According to the method, the adaptability of the model in a heterogeneous edge environment can be remarkably improved, the deployment cost is reduced, and the small-batch and multi-task quick response requirement in a customized production scene is met.

The invention belongs to the field of intelligent vehicle-mounted technology, and particularly relates to a multifunctional vehicle-mounted sensing detection system based on multi-source information fusion, which comprises multi-source information acquisition, high-precision space-time synchronization and heterogeneous data preprocessing. A heterogeneous sensing system with feature level deep fusion, vehicle state dynamic performance evaluation and adaptive decision and risk evaluation capabilities is constructed, and a driving decision is dynamically adjusted through fusion perception and vehicle performance. The vehicle state and dynamic performance evaluation module is constructed by deeply integrating vehicle internal state data acquired by a vehicle-mounted diagnosis system into a perception fusion and decision planning process, and the core contradiction of disjunction of a perception result and a vehicle dynamic performance strategy is effectively solved. The system can evaluate key performance parameters such as power, braking, steering and the like of the vehicle and potential faults of the key performance parameters in real time, and the driving strategy and the safety margin are dynamically adjusted in the self-adaptive decision and risk evaluation module according to the actual physical limitation of the vehicle.

The invention discloses a large model calculation network scheduling method based on task combination automation, and relates to the technical field of data processing, and the method comprises the steps: constructing a task factor flow graph; simulating resource linkage between nodes by using a cascade pulse propagation mechanism, generating a pulse propagation topological graph and a pulse intensity matrix, and constructing a computing force field situation awareness network in combination with a pre-trained pulse graph neural network; predicting the resource matching degree of the task factor and the computing power node by using a preset space-time convolution predictor, and generating an affinity tensor; and constructing a joint strategy space, searching and generating a fusion strategy in the joint strategy space by using a multi-target equalization algorithm, and embedding the fusion strategy into the deep reinforcement learning framework to generate an optimal scheduling strategy. By constructing the computing power field situation awareness network, continuous tracking and awareness of the load evolution process, the resource coupling relation and the performance bottleneck dynamic migration of each computing power node are realized, and the self-adaptive decision-making capability and the resource matching efficiency of a scheduling system in a multi-source heterogeneous environment are improved.

The invention provides a damage assessment method based on heterogeneous double-flow network and multi-strategy adaptive decision fusion. The method aims at solving the technical defects of an existing method in the aspects of dual-time-phase feature alignment, multi-scale perception and fusion strategy self-adaption. The method comprises the following steps: registering and enhancing unmanned aerial vehicle images before and after damage; afterwards, feature extraction is carried out through a heterogeneous double-current feature extraction module, a cross-temporal window alignment mechanism is introduced into a global context awareness stream, strict spatial alignment of double-temporal features is ensured, and capture of multi-scale damage details is enhanced through multi-receptive-field optimization of a local detail enhancement stream; finally, through a multi-strategy self-adaptive decision fusion module, a strategy selector is used for dynamically calculating the weight, weighted decision is carried out on output of the association perception fusion strategy, the explicit change detection strategy and the robust weighted fusion strategy, and finally the damage level is output through a classifier. According to the invention, the accuracy, robustness and adaptive ability of damage assessment in a complex battlefield environment are effectively improved.

The invention relates to the technical field of industrial vehicle intelligence, in particular to an intelligent flexible forklift system based on multi-mode sensing and self-adaptive path planning. The intelligent forklift comprises an intelligent forklift body, a vehicle-mounted integrated control system, a multi-mode sensing and navigation module, a self-adaptive decision-making and path planning module, a cooperative communication and task management module and a cloud scheduling and management platform, and the intelligent forklift body is provided with a pallet fork mechanism with a 3D camera and a force sensor; the vehicle-mounted integrated control system serves as a core to achieve multi-module data interaction and instruction output, the multi-mode sensing and navigation module generates an environment situation map through multi-source data fusion, the self-adaptive decision-making and path planning module completes path planning, obstacle avoidance and pre-verification, the cooperative communication and task management module achieves multi-vehicle cooperation and task disassembly, and the multi-mode sensing and navigation module generates an environment situation map through multi-source data fusion. And the cloud scheduling and management platform realizes global monitoring and management. According to the method, the adaptability to complex scenes and the operation safety are improved, and accurate forking and flexible path planning are realized.

The invention relates to the technical field of image analysis, in particular to a metal composite material surface defect intelligent detection method based on machine vision, which comprises the following steps: acquiring an original image and converting the original image into an observation matrix; decomposing the matrix into a low-rank matrix and a sparse matrix through a robust principal component analysis algorithm; utilizing a nuclear norm constraint low-rank matrix to establish a background statistical model, and utilizing a norm constraint sparse matrix; positioning a defect area in the sparse matrix and extracting a gray level co-occurrence matrix feature vector; performing connected domain analysis on the sparse matrix, calculating geometrical morphology parameters and evaluating a stress concentration coefficient; and establishing a self-adaptive decision model to carry out fusion judgment so as to judge the physical damage. According to the method, through low-rank sparse matrix decoupling and multi-dimensional statistical geometric feature fusion analysis, accurate stripping of weak defects and quantitative evaluation of physical damage attributes under a complex background are realized.

A mine earthquake adaptive positioning method and system based on multi-index fusion discrimination belong to the field of mine earthquake positioning, and the system comprises a data acquisition module, a feature extraction and quality evaluation module, a comprehensive index calculation module and an adaptive decision module. The method comprises the following steps: firstly, carrying out multi-dimensional physical examination on input waveform data, station layout and a speed model by the system, and generating a quantized physical examination report comprising a signal-to-noise ratio, a pickup error, station geometric coverage, model credibility and the like, namely a multi-dimensional feature vector; and then, a built-in rule engine automatically screens out an optimal single method or combined strategy under the current condition from multiple positioning methods according to the report and a demand target set by a user, and outputs a decision result with a clear basis. Through the system and the method, full-process automation from data input to strategy output is realized, and efficient and reliable technical support is provided for mine safety monitoring and disaster early warning.

The invention relates to the technical field of vehicle safety, and provides a remote lane departure early warning method and system based on remote driving, and the method comprises the steps: guiding a fusion network to extract a lane line feature vector through employing a two-way decoder, and generating a vehicle state parameter set through combining with a dynamic modulation Kalman filter; a dual-mode risk quantification framework is applied, the early warning level is judged through a situation self-adaption decision, and an early warning information message is reported; performing time sequence alignment on the comprehensive control instruction set, the driver state data and the environment context data, and generating a control instruction message through optimization of an environment adaptive instruction intention filter; and the vehicle-mounted terminal calculates a man-machine control fusion weight based on the comprehensive risk and the driver state, a second comprehensive control instruction set and a safety deviation correction instruction are fused to generate a bottom layer control signal, and the remote cockpit executes graded early warning and generates a multi-mode feedback signal. According to the invention, prospective risk research and judgment and adaptive control optimization are cooperated, and a man-machine cooperative early warning and closed-loop control mechanism for remote driving is constructed.

The invention discloses a rope net ladder data production scheduling and traceability system based on big data storage, and particularly relates to the field of data scheduling and traceability, comprising the following steps: the system obtains total factor data of equipment, process, quality and the like through a data perception and digital acquisition module and standardizes the total factor data; the integrated storage and quantitative analysis module constructs functions such as efficiency evaluation and quality risk to quantify production indexes; the adaptive decision optimization and scheduling module generates an optimal scheduling scheme and a dynamic rescheduling instruction based on a multi-target comprehensive decision function; the full-link tracing and self-feedback optimization module realizes full-link bidirectional tracing and drives process parameter and model self-feedback optimization; the system realizes the intelligent control of the whole production process, balances the efficiency, quality, cost and delivery target, improves the resource utilization rate and production stability, and is suitable for the precise production scene of structural members such as rope net ladders.

The invention relates to the technical field of intelligent switching, in particular to a multi-mode dynamic coupling intelligent switching method and system based on road condition recognition and a vehicle. The method comprises the steps that gradient, vehicle speed, accelerator and brake signals are collected in real time, and a multi-dimensional feature vector fusing driving habits is constructed after preprocessing; a real-time road condition label is automatically generated through unsupervised clustering, and the clustering effectiveness is verified online by using a contour coefficient. Enabling the road condition label and the current power mode to form a system state, and inputting the system state into a self-learning decision network based on reinforcement learning; and the network updates the state-action value matrix through a reward function in a time sequence difference mode, and outputs a target power mode with the highest accumulated reward. And calculating the motion confidence coefficient of the target mode based on the value matrix, if the motion confidence coefficient exceeds a threshold value, generating a switching instruction, and smoothly adjusting the torque by an execution mechanism through a torque transition algorithm to realize impact-free switching. Automatic road condition recognition, self-adaptive decision making and non-inductive execution are achieved, and smoothness and adaptability are improved.

The invention relates to the technical field of information, and discloses a comprehensive operation supervision system based on an intelligent supply chain. The system comprises a real-time data acquisition module, a multi-modal feature extraction module, a dynamic risk assessment module, a self-adaptive decision generation module and an execution feedback regulation and control module. By constructing a real-time data acquisition and multi-modal feature extraction module, efficient fusion and deep feature extraction of internal and external multi-source heterogeneous data of the supply chain are realized, the sensing sensitivity and accuracy of the system to market fluctuation and emergencies are remarkably improved, and the prediction deviation is reduced from the data source. The dynamic risk assessment module adopts a dual anomaly detection mechanism, various potential risks can be quickly and comprehensively identified, risk level quantification is carried out in combination with a time sequence mode, and accurate and timely risk situation assessment is provided for subsequent decision making.

The invention discloses a traditional Chinese medicine grinding control system, which relates to the technical field of traditional Chinese medicine grinding control and comprises a medicinal material characteristic sensing module, an intelligent calculation analysis module, a dynamic execution adjustment module, a multi-mode optimization control module and a system coordination updating module. The medicinal material characteristic sensing module collects physical and texture characteristic data of medicinal materials; the intelligent calculation and analysis module fuses the medicinal material processing knowledge base and historical grinding process data to generate initial grinding control parameters; the dynamic execution adjustment module adjusts the operation parameters of the grinding mechanism in combination with the real-time grinding state data; the multi-modal optimization control module monitors the deviation between actual quality parameters and target quality parameters of powder, and a self-adaptive decision algorithm is used for optimizing regulation and control parameters of a grinding area; the system coordination updating module collects whole-process data and performance index evaluation and generates a system optimization instruction, and the instruction corresponding module adjusts the rotating speed and pressure parameters of a grinding area of the vertical grinding machine. According to the system, intelligent and precise control over medicinal material grinding is achieved, and the stability and consistency of the grinding quality are improved.

The invention relates to a self-adaptive retrieval enhancement generation system based on multi-dimensional problem features and an implementation method. The system comprises an RAG initialization module, a text processing module, a model initialization module, a vector processing module, a retriever module and an RAG retrieval generation module. The input end of the retriever module receives the text vector of the vector processing module, the output end of the retriever module is connected with the RAG retrieval generation module, a strategy pool of five types of retrieval strategies and the self-adaptive decision module are arranged in the retriever module, and the retriever module is used for determining retrieval strategies through keyword matching and problem length judgment and constructing RAG cue words. And the RAG retrieval generation module is used for calling the retriever module to input the obtained text segment into the large language model to generate an answer after reconstructing the user question. By adopting the method, the retrieval efficiency and the answer generation quality of the RAG system can be improved.

A computer-implemented system for improving radiation dose reconstruction in adaptive real-time radiotherapy, comprising a data acquisition module for capturing treatment data in real time during radiotherapy, a preprocessing module for preprocessing the captured data, a deep-learning dose reconstruction engine for reconstructing the delivered radiation dose distribution using a trained deep-learning model, a dose evaluation and comparison module for comparing the reconstructed dose distribution with a planned dose distribution, and an adaptive decision support module to assist in adjusting radiotherapy treatment parameters based on the comparison results.

The invention discloses an intelligent sports event decision-making system based on causal-driven multi-modal fusion and spatio-temporal dynamic reasoning. The system comprises a multi-modal causal data acquisition and preprocessing module; a causal-oriented multi-modal knowledge graph construction and updating module, wherein the causal-oriented multi-modal knowledge graph construction and updating module is provided with a causal structure learning algorithm and an online learning framework; a space-time perception hybrid agent module; the causal constrained dynamic decision optimization module is provided with a deep reinforcement learning algorithm; and the interpretability analysis and visualization module is used for receiving the decision strategy sent by the space-time perception hybrid agent module, generating decision explanation according to the decision strategy and visually presenting the decision explanation. The objective of the invention is to construct an intelligent system capable of providing high-precision, interpretable and adaptive decision support by integrating causal reasoning, multi-modal learning and reinforcement learning, so as to significantly improve scientificity and effectiveness of sports event analysis and decision.

The invention relates to the technical field of vehicles, in particular to a multi-mode out-of-vehicle interaction method and related equipment based on environment self-adaptation, and the method comprises the steps: obtaining perception information in real time through an environment perception module, the perception information comprising environment conditions, interaction object information and scene semantic understanding information; based on the perception information, utilizing an adaptive decision engine to dynamically analyze an optimized interaction mode and parameters, and generating interaction content; outputting interaction information according to the interaction content generated by the self-adaptive decision engine through a multi-modal output module; the interaction strategy is dynamically adjusted by sensing the environment state in real time, and the safety and the interaction efficiency can be improved.

The invention discloses a high-fidelity robust watermarking method based on deep reinforcement learning. The method is composed of a reinforcement learning module, a watermark encoding and decoding module and a watermark synchronization module in sequence. The reinforcement learning module automatically selects an optimal watermark embedding position in a local area of the image through an adaptive decision-making mechanism of a deep Q network to realize dynamic optimization of an embedding strategy; the watermark encoding and decoding module adopts a convolutional block attention mechanism (CBAM) and a multi-scale feature fusion structure, and realizes high-fidelity embedding and accurate extraction of watermark information while maintaining the visual quality of the image; the watermark synchronization module is based on a U2Net model structure, and completes spatial alignment of a watermark area through mask prediction and affine correction, thereby effectively solving positioning deviation caused by geometric attacks such as rotation, zooming, cutting and the like. The three components work cooperatively, and high watermark extraction precision can be kept under various compression, noise and geometric disturbance conditions. The method has the advantages of imperceptible embedding, high extraction stability, independent training and combined deployment of modules and the like, and can be widely applied to the fields of digital image copyright protection, AIGC (Artificial Intelligence Generation Content) traceability, multimedia authenticity verification and the like.

The invention discloses a load power balance regulation and control method and device based on knowledge graph and random sequence optimization, a medium and a product, and relates to the field of power system dispatching, and the method comprises the steps: constructing a power grid topology knowledge graph according to power grid physical equipment and corresponding topology connection and operation constraints; performing probability distribution fitting on the power output of the renewable energy according to the output data to obtain the output of the renewable energy; obtaining a plurality of typical output scenes; taking the state of the power grid physical equipment after fault isolation as an initial state, and constructing a random Markov decision process; searching in a state space by using an improved heuristic search algorithm by taking an initial state as a starting point, minimizing an accumulated cost function as a target, taking a hard constraint as a constraint condition and taking a topological distance as a heuristic function to obtain an optimal operation sequence; verifying the optimal operation sequence to obtain a final operation sequence; the power of the power grid is regulated and controlled; according to the invention, the fault response speed can be improved, and rapid adaptive decision in a complex scene can be realized.

The unmanned aerial vehicle path planning method based on chaos adaptive particle swarm optimization comprises the following steps: constructing a three-dimensional task scene model, setting flight performance constraint conditions, and constructing a flight path fitness objective function; initializing particle swarm parameters and population positions; calculating an adaptive value of each particle, and updating an individual historical optimal adaptive value / path and a group global optimal adaptive value / path; updating the particle speed and position; updating a current weight parameter based on a dynamic parameter self-adaptive decision-making mechanism; on the basis of a conditional trigger type population variation strategy, the positions of the particles are updated until the number of iterations reaches the maximum value, and an optimal path is output; according to the method, the optimal path is planned based on the chaos adaptive particle swarm algorithm by constructing the flight path fitness objective function, the initial solution of the algorithm is better, environment-population dual perception is realized, the convergence speed of path planning is improved, global exploration and local development are better balanced, and the method has better randomness and robustness.

The invention belongs to the field of multi-room negative pressure gradient control, and discloses a multi-room negative pressure gradient automatic control method and system based on reinforcement learning. An environment model containing multi-dimensional dynamic parameters such as an environment pressure state, an access control state, a fan air volume and a room coupling relation is constructed, so that the method can directly obtain an optimal control strategy through interactive learning on the premise of not needing an accurate physical model. Through the self-adaptive decision-making capability of reinforcement learning, linkage adjustment of the exhaust air volume and the fresh air volume of each room and the power of a total exhaust fan and a fresh air fan of the whole system is achieved, so that the stability of gradients of a corridor-5Pa, a ward-10Pa, a semi-polluted area-15Pa, a polluted area-20Pa and the like is kept, and it is ensured that the pressure difference of all areas is always within the safety range of-7Pa to-2Pa.

The invention relates to the technical field of road construction automation and intelligent control, discloses an intelligent sensing and control road laying automation method and system, and aims to solve the problem of unstable construction quality caused by discontinuous sensing data, poor equipment collaboration and lack of closed-loop feedback in the prior art. The method comprises the steps that operation parameters and material state data of paving and compacting equipment are collected in real time through a multi-mode sensing module; performing space-time alignment and fusion on the multi-source heterogeneous data, and constructing a high-precision three-dimensional digital construction model; and on the basis of a model prediction control algorithm, combining the laying quality prediction model and multi-objective optimization to generate optimal equipment control parameters. According to the scheme, intelligent sensing, self-adaptive decision making and accurate control in the construction process are achieved, the pavement flatness, the compaction uniformity and the construction efficiency are remarkably improved, manual dependence and resource waste are reduced, and the construction safety and the engineering quality traceability are enhanced.

The invention relates to the technical field of automatic control, in particular to a self-learning joint control method and system based on multi-modal feedback and reinforcement learning, and the method comprises the following steps: constructing a multi-modal synchronous sensing array to obtain multi-modal real-time sensing data; performing multi-level feature extraction and deep fusion on the multi-modal real-time sensing data to obtain a high-dimensional state vector; constructing an adaptive decision engine based on deep reinforcement learning to output a joint control instruction; establishing a multi-target composite reward function, and obtaining a composite reward quantitative index of the joint control instruction; introducing an online training optimization mechanism, and performing training optimization on the adaptive decision engine to obtain a joint optimization control instruction; and executing the joint optimization control instruction and obtaining full-state actual joint data, and performing closed-loop iteration on the adaptive decision engine to complete self-learning joint control. According to the invention, high-precision automatic jointing is realized, the jointing quality and efficiency are improved, and the dependence on artificial experience is reduced.

The invention discloses an intelligent analysis and self-adaptive regulation and control system for a glass fiber wiredrawing forming process, which relates to the technical field of high silica glass fiber product processing and comprises a multi-modal data acquisition unit, a process prediction model unit, a reference process model unit, a self-adaptive regulation and control decision unit and a control execution unit. By introducing the prediction model and the self-adaptive decision-making unit, two strong coupling parameters of speed and temperature can be decoupled, collaborative optimization regulation and control are carried out, precise control over the wire drawing process is achieved, the uniformity of the fiber diameter is remarkably improved, the future process trend can be predicted on the basis of real-time data, and the method has good application prospects. Traditional feedback control is improved into predictive feedforward control, fine adjustment is carried out before process parameters are obviously deviated, the rate of finished products is greatly improved, meanwhile, the optimal production process is solidified into a reference model, complex regulation and control logic is precipitated into a decision algorithm, and dependence on artificial experience is eliminated.

The invention relates to the technical field of exercise load electrocardio testing, and discloses an exercise load electrocardio testing system for multi-modal signal processing and intelligent analysis, and the system comprises a multi-modal signal fusion module which is used for constructing a joint data space and integrating multi-source physiological signals; the dynamic feature extraction module is used for extracting features in global and local layers respectively; the self-adaptive decision optimization module is used for constructing a knowledge rule library to generate a feature extraction strategy adaptive to the new scene; and the abnormal state early warning module monitors and triggers an early warning strategy in real time. The system realizes efficient fusion and feature extraction of multi-modal signals through operations of synchronous acquisition, multi-scale decomposition, layered screening and the like, can adaptively optimize strategies according to different test scenes, monitors abnormities in real time and gives an early warning, improves the accuracy, anti-interference performance, adaptability and safety of exercise load electrocardio tests, and has a wide application prospect. The method is suitable for sports medicine and health monitoring fields.

The invention discloses an Internet of Vehicles cooperative unloading method based on hybrid action deep reinforcement learning, and relates to the technical field of Internet of Vehicles and mobile edge computing. In order to solve the technical problem of joint optimization of unloading target selection and resource allocation, the problem is modeled as a Markov decision process, and a mixed action soft actor-commentator (HA-SAC) algorithm is provided for solving; the core is that a multi-head strategy network outputs discrete actions and continuous actions at the same time in a decision period. A centralized training and decentralized execution architecture is adopted, and the value network receiving the global state guides the intelligent agent which only depends on the local state to make decisions to learn; compared with the prior art, the method has the advantages that the precision loss caused by action space discretization is avoided, the task completion time delay is obviously reduced, and the robustness and the adaptive decision-making capability of the system in a highly dynamic network environment are enhanced.

The invention discloses an FISTA plug-and-play sparse radar imaging method based on adaptive regulation and control, which combines FISTA with deep learning prior, introduces reinforcement learning to carry out adaptive parameter regulation and control, and aims to significantly improve the automation degree, convergence rate and cross-scene robustness of an imaging process. The method is especially suitable for processing challenges of coexistence of different sampling rates, noise levels and scene complexity, and can effectively solve the problems that a traditional method depends on manual parameter adjustment, the calculation overhead is high, and the generalization ability is weak. According to the method, selection modeling of key hyper-parameters (such as step length, de-noising intensity and momentum coefficient) and a stop criterion is used as a Markov decision process, and an intelligent agent makes a decision adaptively according to a current reconstruction state, so that the problem is solved, and an advanced solution is provided for efficient and reliable radar imaging processing.

The invention discloses an unmanned aerial vehicle autonomous strike decision self-adaptive adjustment system based on multi-source information fusion, and particularly relates to the technical field of unmanned aerial vehicle autonomous control and self-adaptive decision, the system comprises an information fusion and decision inner ring and a meta-cognitive self-adaptive adjustment outer ring, the inner ring executes multi-source information fusion and generates a strike decision, and the outer ring executes a meta-cognitive self-adaptive adjustment. And the outer ring evaluates the environment and information quality in real time, drives the adjustment strategy generator based on reinforcement learning by comparing the expected efficiency and the actual efficiency of the decision, and dynamically adjusts the fusion weight and the decision parameter of the inner ring or switches the decision model. The problems that an existing system is poor in environmental adaptability and cannot achieve online self-optimization are solved, autonomous, closed-loop and continuous optimization of decision-making performance in a dynamic battlefield environment is achieved, and the autonomous combat effectiveness, the environmental adaptability and the task robustness of the unmanned aerial vehicle are remarkably improved.

The invention discloses a distributed AI training-oriented RDMA transmission mode intelligent selection method, and the method comprises the steps: obtaining multi-dimensional data features in a deep learning training process, the multi-dimensional data features comprising a tensor structure feature, a data access mode feature, a data timeliness feature, a data dependency feature and a data change rate feature; identifying a current training stage, wherein the training stage comprises a forward propagation stage, a back propagation stage and a parameter updating stage; based on the multi-dimensional data features and the training stage, an optimal RDMA transmission mode matched with the current training scene is selected through an adaptive decision matrix, and the optimal RDMA transmission mode comprises one or more combinations of RDMA Write operation, RDMA Read operation, Send / Receive operation and Atomic operation; and switching the RDMA transmission mode in real time according to the dynamic change of a network environment, ensuring that a training data flow is not interrupted in a switching process by adopting a progressive flow migration strategy, and optimizing a subsequent mode selection decision through a historical performance learning mechanism.