146 results about "Collaborative strategy" patented technology

The invention discloses a multi-source knowledge-enhanced big language model question and answer method, device and equipment and a medium, and the method comprises the steps: inputting a user question into a big language model, and obtaining an initial answer and a cooperation strategy deduced by the big language model for the user question; after taking the initial answer as a current candidate answer, extracting a target knowledge source type in a collaborative strategy and generating a corresponding calling expression; according to the calling expressions, target knowledge sources matched with the target knowledge source types are called respectively, and an enhanced retrieval document set is obtained; re-inputting the user question and the enhanced retrieval document set into the large language model to obtain a new answer and a collaboration strategy; and taking the new answer as the current candidate answer, repeating the operation of extracting the target knowledge source type, and determining a target answer from all the candidate answers as a feedback result when an iteration ending condition is met. According to the technical scheme, the question and answer accuracy can be optimized through iterative retrieval and knowledge enhancement.

The invention relates to the technical field of intelligent agricultural robots, and provides a multi-agent collaborative anti-collision picking method based on digital twinning and deep reinforcement learning, which comprises the following steps: constructing a digital twinning model of a picking scene, and generating environmental geometric parameters, agent kinetic parameters and fruit position parameters through three-dimensional point cloud reconstruction; acquiring environment state data in real time and inputting the environment state data into the digital twin model for space-time alignment processing to generate synchronous state data; based on the synchronous data, a collaborative strategy containing a collision avoidance priority matrix, a path planning sequence and a task allocation weight is generated through a deep reinforcement learning network; an action instruction set is generated according to the strategy, and multiple agents are controlled to execute a picking task after virtual-physical space bidirectional verification of the digital twin model. According to the invention, efficient collision avoidance and accurate picking of multiple agents in a dynamic environment can be realized, and the picking efficiency, safety and system robustness are improved.

The invention relates to a multi-agent cooperation strategy generation method, device and equipment and a medium, and the method comprises the steps: separating acoustic spectrum features and text semantic features of conference voice through environment perception processing, solving a cross-modal information conflict problem, and generating an accurate semantic understanding result; identifying the essence of the problem based on task analysis, associating the responsibility field, and constructing a classifiable problem point set; calling an agent capability library to dynamically match problem requirements, and generating a candidate agent list; quantifying a problem influence range and a decision time limit through weighted emergency scores, and generating a priority-sorted agent sequence; screening and confirming a core problem point and a primary agent; and finally, generating an executable cooperation scheme through multi-agent collaborative optimization. According to the method, the problems of incomplete feature extraction, task allocation delay and resource conflict in the prior art are solved, and the operability and decision-making efficiency of a cooperation strategy are remarkably improved.

The invention relates to the technical field of data processing, in particular to a data fusion intelligent equipment linkage management and control system based on a large model, and the system comprises a data collection module which outputs standardized data carrying privacy grading identifiers; the data processing module outputs an optimization algorithm identifier and a collaborative strategy identifier which are synchronously bound; the privacy algorithm fusion scheduling module generates atomization calculation graph data and strategy audit logs fusing privacy operation; the multiple optimization resource module executes calculation graph data and drives a parallel simulation environment rehearsal resource scheme, and the double-closed-loop feedback module is activated when an arbiter continuously detects that deviation exceeds the standard for three times; the safety instruction generation module fuses the output result and the audit log to generate a digital signature equipment control instruction; and the double-closed-loop feedback module updates knowledge base weight parameters through a data feedback loop, and controls the feedback loop to adjust a resource allocation threshold in a grading manner according to simulation confidence, so that dynamic balance between resource efficiency and privacy intensity is realized.

The embodiment of the invention discloses a display panel production scheduling method and system based on agent collaboration, and the method comprises the steps: carrying out the real-time state collaborative updating operation of a plurality of agents in a display panel production system, and obtaining a collaborative state set containing the identification information of the agents and the description of a corresponding production related state; generating an inter-agent cooperation strategy set containing information interaction rules and task allocation priority description based on the set; calling a pre-configured negotiation mechanism to perform conflict detection and negotiation processing on the cooperation strategy set to obtain an optimized cooperation strategy after conflict resolution; and finally, dynamic scheduling operation of display panel production is executed according to the optimized cooperation strategy, and a production scheduling instruction set containing a process execution sequence and an equipment allocation scheme is generated. According to the embodiment of the invention, through intelligent agent cooperation and dynamic scheduling, optimal scheduling of display panel production is realized, and the production efficiency and quality are improved.

The invention provides a consensus decision question answering system based on multi-AI agent game, and relates to the technical field of artificial intelligence. The system comprises a multi-domain information aggregation module, an interaction effect deduction module, a strategy fusion calibration unit, a distributed behavior adaptive mechanism and an aggregation strategy discrimination module. The multi-domain information gathering module is used for unifying multi-source strategy information and environment situation data, the interaction effect deduction module is used for analyzing and quantifying the mutual influence relation of strategies between intelligent agents, and the strategy fusion calibration unit generates correction suggestions based on a game deduction and optimization method. The distributed behavior self-adaptive mechanism is used for locally and progressively executing a correction path in an intelligent agent; and the aggregation strategy judgment module dynamically evaluates the overall strategy state. The multi-agent consensus decision-making question-answering method realizes consensus decision-making question-answering of multiple agents in a complex environment, can effectively identify and correct non-collaborative strategy deviation, and improves the coordination, stability and immunity of a system.

The invention discloses a multi-agent cooperation system construction method, a medium and equipment, and the method comprises the steps: carrying out the task modeling of a target business scene, and constructing a task dependence graph; initial cooperation strength weights are set for the atomic tasks with the cooperation relationship, and multi-stage cooperation training courses from easy to difficult are generated based on the cooperation complexity of the atomic tasks; the intelligent agent is controlled to execute a task in a training course, interaction behavior data is collected, the overall task completion efficiency is calculated, the cooperation weight is dynamically updated, and interaction data and the updated weight are input into a reinforcement learning model to iteratively optimize a cooperation strategy; and finally, solidifying the converged cooperation strategy into the constructed multi-agent cooperation system. According to the invention, through the combination of course learning and reinforcement learning, the cooperation efficiency and robustness of the system in a complex business scene can be significantly improved.

Disclosed are a force guidance telerobotic system and control method based on a dual-arm collaborative potential field. A real-time pose of a tool center point of each of the two robotic arms is obtained using a robotic arm kinematic model, and checking is performed to determine whether a shortest distance dp-s,i between a target object and a workspace boundary of each of the two robotic arms is lower than a threshold Ds; a dual-arm symmetric collaboration strategy will be adopted when higher than the threshold; a single-arm primary collaboration strategy will be adopted when lower than the threshold; and a coordination factor δi of each of the robotic arms is determined according to the collaboration strategy; the dual-arm collaborative potential field is constructed according to the collaboration factor, a distance between the target object and the tool center point and a position of obstacle.

The invention discloses a multi-task collaborative incremental webpage data acquisition method and system, and particularly relates to the technical field of webpage data acquisition, which comprises the following steps: performing primary evaluation on a target webpage to calculate a structural uncertainty score, judging whether to enter deep evaluation to generate a page structure consistency index and a content update index, and if yes, determining that the target webpage is consistent with the content update index; outputting an acquisition scheduling coefficient through the acquisition scheduling prediction model, and selecting a structure reconstruction mark tracking or content difference snapshot comparison path; task strategy types are classified in combination with the combination relation of the two indexes, and multi-task collaborative collection is achieved according to the classification result and a preset collaborative strategy; according to the method, webpage structure and content changes can be dynamically perceived, a structure consistency index and a content updating index are generated, and the judgment precision before collection is improved; an adaptive path is selected according to the collection scheduling coefficient, and the collection stability in a complex webpage environment is enhanced; and in combination with a classification result and a coordination strategy, resource allocation and time scheduling are optimized, and efficient coordination among multiple tasks is realized.

The invention discloses a humanoid robot motion control reinforcement learning method. The method comprises the following steps: constructing an intelligent agent architecture comprising a sensing module, a reinforcement learning module, a decision-making module and a communication module; hierarchical action control of generating tactical-level sub-targets at a high layer and generating specific action instructions at a bottom layer is realized through a hierarchical reinforcement learning model; in an adversarial environment, adversarial behaviors are predicted through an opponent modeling network, a reward function is embedded, and a shared attention mechanism is adopted to optimize a collaborative strategy; during strategy exploration and parameter updating, strategy entropy regularization items are superposed, and the parameters are updated on the basis of experience playback buffer areas; signals are asynchronously broadcasted through the communication module, and the receiver dynamically adjusts the strategy. According to the method, the universality and adaptability of the humanoid robot system are enhanced, the learning efficiency and strategy optimization capability of the humanoid robot intelligent agent are improved, the performance of the humanoid robot in a non-cooperative environment is improved, the robustness and stability of the system are enhanced, and the method can adapt to a complex dynamic environment.

The invention discloses a multi-platform-based advertisement creativity and putting strategy intelligent generation system, equipment and medium, multi-source data are integrated through a cross-platform creativity intelligent generation module to generate creativity materials adaptive to different platforms, and a cross-platform collaborative strategy optimization module is used for quantifying a platform collaborative effect and dynamically allocating budget. An advertisement creative life cycle is managed by means of a real-time dynamic optimization module, a strategy is adjusted based on multi-dimensional feedback, and finally, output of each module is fused through a unified decision engine to form a unified decision. The system realizes full-process intelligentization of advertisement originality and putting strategies, quantifies a cross-platform synergistic effect to improve budget distribution efficiency, constructs an advertisement originality life cycle dynamic management mechanism, realizes strategy real-time evolution through multi-dimensional feedback, and visualizes full-link data in combination with a unified decision engine, so that the advertisement putting return on investment is remarkably improved, and the advertisement putting efficiency is improved. The creative manufacturing cost is reduced, the strategy adjustment efficiency is improved, and an efficient and accurate multi-platform putting solution is provided for advertisers.

The invention relates to the technical field of privacy protection of networked automobiles, and discloses a privacy protection distributed federated learning collaborative strategy for networked automobiles, which comprises a privacy protection system of networked automobiles, and the privacy protection system of networked automobiles comprises an application scene layer, a core technology layer, an edge computing layer and a data and equipment layer. According to the privacy protection distributed federal learning cooperation strategy for the networked automobile, real-time and multi-dimensional perception of a traffic environment is realized through integration of V2X communication, a road condition space-time encoder, a stream data processing engine and other technologies, accurate data support is provided for intelligent driving decision making, and the privacy protection distributed federal learning cooperation strategy for the networked automobile is provided for the intelligent driving decision making. Through a state division domain strategy module and a time-space domain adaptive division technology, resource allocation and task scheduling can be dynamically adjusted according to real-time road conditions and network conditions, the response speed and efficiency of the system are improved, and through technologies of a privacy calculation pool module, dynamic privacy budget allocation, a PKI certificate service center and the like, real-time dynamic privacy budget allocation can be realized. And strict privacy protection is realized during data sharing and federal learning.

The invention relates to the technical field of large model driven reasoning, and particularly discloses a multi-agent interaction intention understanding and cooperative control method based on a large model, and the method comprises the following steps: S1, environment and multi-source heterogeneous interaction information perception; s2, large model driven hierarchical intention understanding; s3, generating an intention-guided collaborative strategy; s4, action execution and closed-loop online optimization are carried out; through full-link technical innovation, the intention understanding precision, cooperative control efficiency and scene adaptation capability of the multi-agent system are remarkably improved, core technical support is provided for multi-agent cooperative application in a complex scene, and the method has extremely high engineering application value and industrial popularization potential.

The invention belongs to the technical field of power grids, and discloses a collaborative optimization method, system, equipment and medium based on a virtual power plant operator and a producer, and the method comprises the following steps: S1, building a power-carbon collaborative management framework of the virtual power plant operator and the producer; s2, constructing a double-layer game model based on the power-carbon collaborative management framework; the upper layer is an operator, the lower layer is a consumer, the operator calculates the node marginal electricity price and the node carbon intensity through the optimal power flow and issues the node marginal electricity price and the node carbon intensity to the consumer, and the consumer optimizes scheduling based on the node marginal electricity price and the node carbon intensity and submits an electricity purchasing and selling curve; and S3, solving the double-layer game model by using a distributed iterative algorithm of a penalty function, and realizing adaptive optimization of model solving by introducing a penalty term to obtain an optimal power-carbon collaborative strategy.

The invention relates to a thermal power plant multi-agent collaborative operation platform construction method, which comprises the following steps: collecting time sequence and static data of a whole plant, and carrying out data preprocessing; sending the data into a digital twinborn body, performing data feature fusion by using a neural network model, generating a high-dimensional feature vector, and sending the high-dimensional feature vector into a digital twinborn model for prediction; a specialized agent is constructed, and training is carried out in the digital twinborn body; a plurality of agents perform multi-step time sequence prediction by using a built-in prediction model to obtain a local optimization decision; constructing a collaborative brain based on a large model, receiving a local optimization decision, and generating a global state code; constructing a decision planner, receiving the global state code and outputting a collaborative strategy instruction; and integrating the digital twinborn body, the specialized agent, the collaborative brain and the domain knowledge base, constructing a unified collaborative operation platform, and carrying out iterative optimization on the digital twinborn prediction model and the collaborative brain. Compared with the prior art, the method has the advantage of high cooperative processing efficiency.

The invention discloses a two-stage multi-agent cooperation strategy based on exploration reward molding, and belongs to the field of multi-agent reinforcement learning. Trajectory data (including environment states, rewards, rewards and actions) generated by interaction of the intelligent agent and the environment are stored in an experience buffer pool and are updated and maintained through increment. Subsequently, randomly sampling an environment state and a corresponding return from the experience pool, and constructing a conditional diffusion model by using the return as a condition to generate a high-return target state; thirdly, global environment states of different time steps in different trajectories are sampled, time structure mapping is learned, and states with similar time are mapped to hidden states with similar geometric space; according to the method, a double-end Q network is adopted, an exploration strategy is decoupled into a target exploration strategy and a behavior exploration strategy, reward functions corresponding to two stages respectively act on a decision-making network of an intelligent agent, and more effective exploration and collaboration are achieved.

The invention discloses a production operation benchmarking system and method based on multi-source data analysis, and belongs to the technical field of production operation benchmarking, and the system comprises a multi-source confrontation fusion module, a production situation awareness map module, a self-adaptive diagnosis analysis module, a cross-domain collaborative deduction strategy module, a hidden variable benchmarking analysis module and a cross-scene migration module. According to the method, enterprise resources are quantized into a matrix, cross-domain resource allocation is optimized in combination with a collaborative gain function, collaborative gains are measured through a nuclear norm and a Frobenius norm, multi-target weighting is combined, priorities of different fields are balanced, the global optimality of a collaborative strategy is improved, the enterprise resource state is synchronized to a digital twinborn model, and the collaborative resource allocation efficiency is improved. A strategy execution process is simulated, burst interference is injected, strategy stability is verified, a cross-domain cooperation efficiency comprehensive value is calculated through a four-flow state, a single-domain independent optimization mean value is compared, and cooperation gain is quantified.

The invention provides an exoskeleton system assistance control method and device based on a man-machine co-fusion strategy, and the method comprises the steps: collecting basic motion data of a human body and an exoskeleton system through various types of equipment, and initializing the motion state of the exoskeleton system; constructing a man-machine co-fusion kinetic model based on the basic motion data; on the basis of a man-machine co-fusion kinetic model, processing the collected real-time data through an EKF algorithm, and estimating various dynamic parameters of a human body and an exoskeleton system; determining a power-assisted control strategy of the exoskeleton system in combination with the dynamic parameters and a man-machine co-fusion kinetic model, and adjusting the power-assisted control strategy according to the personalized parameters of the user; and controlling the exoskeleton system to execute the adjusted power-assisted control strategy. According to the method, the man-machine collaborative motion intention prediction algorithm is adopted for power assisting control, the power assisting accuracy and flexibility of the exoskeleton system can be improved, and the method can dynamically adapt to individual differences of users.

A big and small model collaborative semantic rewriting system based on complexity sensing comprises a user question receiving module, a context extraction module, a complexity evaluation module, a collaborative strategy module and an output rewriting module, and the user question receiving module is used for receiving questions of a current user and starting a semantic rewriting process. The context extraction module is used for integrating historical dialogue contexts and extracting key entities, the complexity evaluation module is used for evaluating user question semantic complexity and driving strategy decision, the collaborative strategy module is used for dynamically scheduling collaborative modes of small models and large models and optimizing resource allocation, and the output rewriting module is used for generating final rewritten questions. According to the large and small model collaborative semantic rewriting system based on complexity induction, a semantic complexity evaluation algorithm based on machine learning is put forward to evaluate semantic complexity, and a model collaborative strategy algorithm based on a dynamic threshold is put forward to cooperatively process tasks of all levels.

The invention discloses an optimization control method for a power system, and relates to the technical field of power control, and the method comprises the steps: constructing a multi-energy state space, and building a layered optimization architecture based on the multi-energy state space; the decision-making layer generates a multi-energy balance scheduling instruction through a multi-target collaborative optimization algorithm to realize multi-target collaborative optimization of the electric heating gas system; the execution layer comprises a plurality of energy subsystem control modules, and collaborative strategy optimization is carried out through a distributed reinforcement learning algorithm; positioning a disturbance source of topology mutation through a causal reasoning algorithm, and generating a reconstruction instruction; training a control strategy of each energy subsystem by adopting a distributed learning framework, and executing a distributed cooperative control action according to the reconstruction instruction; the network connectivity is judged through the communication state of the energy subsystem, and if the communication is normal, collaborative decision making is carried out through a distributed learning framework; and if the communication is interrupted, switching to an edge node offline autonomous mode.

The invention discloses a virtual-real fusion-based complex task collaborative twinborn deduction method, which comprises the following steps of: firstly, establishing a multi-dimensional collaborative relation model between a task and equipment, and constructing a scheduling modeling basis from three aspects of equipment topology, resource mapping and task dependence; secondly, constructing a task collaborative deduction and optimization model based on a dynamic game, simulating a state evolution process of multiple tasks in a virtual space, and performing path optimization by taking a system efficiency function as a target; and finally, fusing an annealing mechanism and a virtual gradient perturbation design collaborative strategy optimization method, guiding a path adjustment direction based on efficiency feedback, and realizing multi-round iterative evolution of a task strategy. According to the method, the global optimization capability of the scheduling strategy and the stability of system operation are effectively improved, and the method has wide application prospects in automobile part flexible production task scheduling and collaborative optimization scenes.

The invention discloses a multi-hospital first-aid collaborative strategy recommendation method and system, and relates to the technical field of pre-hospital first-aid collaboration.The method comprises the steps that spare medical resource information is acquired in real time, and an initial collaborative strategy set is acquired based on the spare medical resource information according to department and hospital positions and basic first-aid configuration; acquiring historical medical resource use information, and constructing a medical resource candidate influence relationship based on the medical resource use frequency and the importance degree according to the historical medical resource use information; and responding to the first-aid information, matching an initial collaborative strategy according to the first-aid information, generating an auxiliary collaborative strategy according to the medical resource candidate influence relationship, and executing collaborative strategy recommendation according to the initial collaborative strategy and the auxiliary collaborative strategy. The method has the beneficial effects that the medical resource scheduling efficiency and the anti-risk capability in the emergency treatment process are improved.

The invention provides an optimization method and system suitable for scheduling of a source-network-load-storage integrated base. The method comprises the following steps: S1, updating and predicting information; s2, constructing an optimization model and embedding a collaborative strategy; s3, carrying out optimization solution; and S4, dispatching instruction issuing. The method focuses on the power optimization scheduling problem of a source network load storage integrated energy base, firstly, an optimization model with the operation cost minimization as the target is constructed, and the core of the optimization model is that a mixed operation mechanism and dynamic constraints of advanced adiabatic compressed air energy storage and batteries under the principle of main and auxiliary cooperation are depicted finely. Then, in order to effectively cope with uncertainty of new energy output, a model prediction control framework is introduced to carry out rolling optimization and closed-loop correction on a scheduling plan, and robustness of a strategy is improved. The strategy provided by the invention provides an effective solution for realizing economic and reliable operation of the integrated base at the high-proportion renewable energy source network load.

The invention discloses a distributed power supply cooperative control method based on transient dynamic characteristic adaptive driving, which adopts a multi-dimensional transient performance coupling evaluation function, unifies multiple targets including frequency, voltage, power angle and operation economy into a global optimization target, and fundamentally solves the defect of single target in the prior art. A collaborative strategy network based on dynamic context awareness is adopted, and a time sequence processing technology based on an attention mechanism is utilized, so that an intelligent agent can extract key transient dynamic characteristics from a local observation sequence, and the limitation of no memory and non-self-adaption in the prior art is broken through; according to the method, a cooperative control flow of a cooperative control strategy is provided, a high-performance learning type strategy is combined with a high-reliability safety monitor, the problem that safety and performance are difficult to be compatible is solved, fundamental transformation of distributed power supply control from passive, local and static rule-based modes is achieved, and the transient stability margin of a novel power system is remarkably improved.

The invention discloses an artificial intelligence platform computing power resource scheduling management method and system, and relates to the technical field of artificial intelligence computing power scheduling, and the method comprises the specific steps: hardware topology collection traverses heterogeneous resources, and generates and updates a graph; the collaborative strategy decision is based on the atlas, and optimal precision and parallel combination are screened; load balancing deploying split tasks, matching hardware capability and planning a transmission path; executing and monitoring multi-dimensional real-time tracking; the quantization deviation is dynamically adapted and adjusted, and the strategy is timely optimized; according to the invention, through hardware topology acquisition and collaborative strategy decision, a computing power resource optimal configuration scheme is formed, and the training task starting efficiency and stability are improved; load balancing deployment and dynamic adaptation adjustment are combined, resource balancing allocation and whole-process dynamic optimization are achieved, the problems of resource imbalance and response lag are effectively solved, the computing power utilization efficiency is improved to the maximum extent, and efficient and stable training is guaranteed.

The invention relates to the technical field of intelligent equipment interaction, and discloses a multi-screen cooperative control method and system based on a USB. The method comprises the following steps: acquiring task characteristics of each equipment screen, and analyzing screen vision composition according to the task characteristics to determine a task type; if the task type is a high-real-time category and the transmission link performance is poor, allocating a high-priority mark to the task and adjusting transmission resources; monitoring the execution state of the task with the high priority mark to determine a supplementary instruction; fusing the supplementary instruction and context sensing data to generate a final collaborative instruction set; and allocating and executing the collaborative instruction set, and performing feedback optimization by comparing an execution result with a preset task state. According to the method, by dynamically sensing the task and adaptively adjusting the instruction, the problem that in the prior art, the efficiency is not high due to the fact that a collaborative strategy is rigid is solved.

The invention discloses a large model dynamic optimization method and system for edge-cloud collaborative reasoning, relates to the technical field of large model dynamic optimization, and solves the technical problems that the prior art lacks flexibility and adaptivity and is difficult to effectively deal with complex and changeable edge application scenes. The method comprises the following steps: S1, collecting and analyzing various key information of edge equipment, a cloud end, a large model to be reasoned and a reasoning task in real time; s2, hierarchically dividing the large model to be reasoned, identifying a plurality of layers or sub-modules which can be used as model unloading points, and dynamically determining an optimal unloading point and an edge-cloud collaborative execution strategy of the large model by utilizing an intelligent optimization algorithm based on the information collected in the step S1; s3, executing a reasoning task according to the dynamic unloading and cooperation strategy generated in the step S2, and converging a final result; through the decision module based on deep multi-agent reinforcement learning (DMARL), the optimal unloading point and the cooperation strategy of the large model can be intelligently selected in real time.

The invention relates to the technical field of distributed file management, and discloses a college cross-campus distributed file collaborative management system. The system comprises a file operation sensing module, a collaborative strategy decision module and a distributed execution engine module. The file operation sensing module is deployed at each campus file server node, captures user editing operation and converts the user editing operation into an operation feature vector set with a timestamp; the collaborative strategy decision-making module generates a conflict probability matrix by analyzing the operation feature vector, and generates a transmission path scheme and a real-time synchronization instruction in combination with a network state and load data; and the distributed execution engine module is responsible for analyzing the instruction, adjusting transmission parameters, processing conflicts and generating a global audit tracking report. According to the system, refined perception, conflict pre-judgment, intelligent transmission, dynamic synchronization and comprehensive auditing of cross-campus file operation are realized, the efficiency and reliability of college cross-campus file collaborative management are improved, and the system is suitable for a multi-campus college file collaborative management scene.

The invention discloses a heterogeneous unmanned aerial vehicle-unmanned vehicle cluster task allocation method and a related device, and relates to the technical field of unmanned aerial vehicles, and the method comprises the steps: constructing an intelligent agent capability matrix and a task demand matrix, and calculating the task confidence; establishing an option-comment framework by using a semi-Markov decision process; performing spatial clustering on the task target points by using a K-means + + algorithm to obtain a first clustering result; adjusting the distribution of the unmanned aerial vehicle / unmanned vehicle to a cluster according to the task confidence, and obtaining a second clustering result; calculating path connectivity and generating dense reward feedback; and determining an option set, training a multi-agent cooperation strategy based on reinforcement learning, and outputting a task allocation result. According to the invention, the generalization ability and adaptability of task allocation can be improved.

The invention relates to the technical field of adaptive control, and discloses a heterogeneous energy resource multi-agent aggregation and adaptive control method, which comprises the following steps of: firstly, acquiring state information of energy equipment in real time through a sensor; based on the equipment state information, a multi-objective optimization problem is constructed, and the optimization objective is to maximize the comprehensive power generation efficiency of the system, minimize the operation cost and maintain load balance; according to the method, through a Nash equilibrium model based on the game theory, a collaboration strategy between agents is constructed, the agents are adjusted according to local information, and finally global optimal power output is achieved; and then performing adaptive control on each device, processing external disturbance by using a robust control algorithm, and enabling the system to adjust device operation parameters in real time when the environment changes. By introducing a self-adaptive dynamic optimization strategy based on an augmented Lagrangian multiplier method, real-time optimal scheduling of a multi-source system under a dynamic constraint condition is realized, and the system can still keep stable operation and global optimal control effects in a complex disturbance environment.