132 results about "Decision problem" patented technology

The invention discloses an intelligent target allocation method and system based on deep reinforcement learning, particularly relates to the field of multi-agent systems, and is used for solving the real-time decision problem under multi-dimensional constraints in a dynamic confrontation simulation scene. Time, space and resource constraints are decomposed into independent models, and the decision complexity of a multi-agent system in a dynamic confrontation scene is reduced; by means of adaptive Monte Carlo sampling, prediction perception path planning and dynamic threshold value resource allocation, diversified high-quality candidate schemes are quickly generated; when the environment changes, the multi-layer influence propagation network and the spatial topological relation graph are used for positioning an affected agent, and scheme fragments are generated through local adjustment, so that global re-calculation is avoided, the overhead is reduced, and real-time response is ensured; meanwhile, an efficient and reliable optimal solution is screened out through resource delay coupling and path efficiency comprehensive evaluation in combination with multi-objective optimization and task consistency verification.

The invention discloses an information scanning and penetration test path planning method in a Windows domain based on reinforcement learning. The method comprises the following steps: carrying out information detection on a target Windows domain environment, surveying and mapping a network topology of the target Windows domain environment, establishing connection with a target machine by utilizing protocols such as LDAP (Lightweight Directory Access Point) and the like, and collecting related information in the domain; the method comprises the following steps of: collecting, sorting and abstracting detected related information, solving a multi-stage sequence decision problem by utilizing a Markov decision process, realizing environment simulation, constructing a reinforcement learning agent, generating a penetration test path by performing interactive training with a simulation environment, and performing adjustment according to a reward strategy so as to obtain a penetration test result. And searching an optimal penetration test path by accumulating rewards, constructing a Windows domain penetration test model, and finally generating the optimal penetration test path. According to the method, collection of information in the Windows domain and planning of the penetration test path are realized, and finally, the environmental safety of the Windows domain is evaluated and maintained.

The invention discloses an edge cloud cooperative control method for a multi-axis servo system, and provides an optimized edge cloud cooperative control framework combining the advantages of edge computing and cloud computing aiming at the condition that a plurality of motion axes in the multi-axis cooperative system track different target trajectories during cooperative work. Considering the condition that the state and the control sequence are transmitted through a limited shared channel, an estimated state, an estimated covariance, cache content and a future instruction are further constructed as a system state based on an enhanced Q backpack method, a steady-state Markov decision problem is constructed, and the decision of each step is converted into a packet backpack problem of an estimated Q value, so that the system performance is improved. Through uplink of a scheduling state and downlink of a control sequence, the overall performance of the system is maximized under the constraint that the number of distributable time slots is limited.

The invention relates to a DDRL-based aircraft mission offloading method in a low earth orbit satellite network. The method comprises the following steps: (1) constructing an LEO low earth orbit satellite network system into an LEO satellite mobile edge computing network system; (2) modeling an LEO low earth orbit satellite network system utility maximization problem into a joint decision problem for optimizing task unloading and resource allocation; and (3) converting a joint decision problem into a Markov decision process MDP, learning an optimal task unloading decision by using a dual depth Q network algorithm DDQN under a dual depth reinforcement learning DDRL framework, and obtaining an optimal resource allocation strategy by using a time difference triple strategy gradient algorithm TD3PG on this basis. Simulation experiments show that compared with a reference algorithm, the scheme can effectively access and cooperatively compute concurrent tasks, and has better convergence and superiority under different environment variables.

The invention provides a bridge group maintenance priority dynamic decision-making method and device based on reinforcement learning, and relates to the technical field of bridge intelligent maintenance. The method comprises the following steps: constructing a topological structure of a bridge network and a road; defining a state space, a maintenance action space and a state transition matrix of the bridge; defining a reliability index corresponding to the state of the bridge, and designing a comprehensive reward function based on maintenance cost, asset risk and traffic network capacity loss risk based on the topological structure; constructing a bridge maintenance decision problem; the method comprises the following steps of: describing a bridge maintenance decision problem as a Markov decision process, establishing a pointer network strategy model by adopting a pointer network, and training the pointer network strategy model by adopting an Actor-Critic algorithm to obtain a maintenance decision model based on reinforcement learning; and training the maintenance decision model based on reinforcement learning until convergence, and outputting a bridge maintenance action sequence under limited constraints. By adopting the method, the limitation problem of traditional single bridge assessment can be solved.

The embodiment of the invention discloses a human and AI collaborative large-scale group decision-making method and equipment based on a trusted network. The method comprises the following steps: constructing a decision-making problem; performing initial evaluation on the alternative scheme through a human decision maker set and an AI decision maker set to obtain initial evaluation information; constructing a human and AI trusted network; establishing a continuously learned human and AI consensus reaching algorithm, and dynamically updating the human and AI trust network and the initial evaluation information through the human and AI consensus reaching algorithm to obtain final evaluation information; calculating the weight of each attribute in the alternative scheme and the weight of each decision maker according to the final evaluation information; the final score of each alternative scheme is calculated, and a final decision is generated; the decision-making response speed, precision and decision-making consistency can be effectively improved, and the method is suitable for the complex decision-making field.

The invention discloses a low-carbon economic operation optimization method for an electricity-hydrogen-heat comprehensive energy system, and belongs to the crossing field of intelligent energy systems and artificial intelligence, and the method comprises the following steps: 1, building a low-carbon economic dual-target operation optimization problem model of the electricity-hydrogen-heat comprehensive energy system; 2, constructing a dual-objective operation optimization problem solving framework with a third-generation non-dominated sorting genetic algorithm as a core; 3, modeling a population evolution decision problem in the genetic algorithm into a Markov decision process, and constructing a deep reinforcement learning agent and population evolution environment interaction mechanism; 4, training the intelligent agent by adopting a double-depth Q network algorithm, and outputting a genetic action according to an environment state; 5, obtaining a new filial generation in combination with a selection strategy associated with the reference points and genetic actions output by the intelligent agent; the step 4 and the step 5 are iteratively evolved until the maximum iteration generation number is reached, and the operation cost can be reduced under the same carbon emission condition.

The invention discloses a polygonal and quadrilateral mesh automatic subdivision method based on deep reinforcement learning, relates to the field of computational geometry and mesh generation, and constructs an automatic subdivision framework fusing geometric priori knowledge and a data driving strategy. According to the method, nine basic topology filling templates covering triangles to hexagons are predefined as discrete action spaces, and continuous geometric segmentation is converted into a sequence decision problem; learning a mapping relation between the polygon state characteristics and the optimal template selection strategy by using a deep Q network, and balancing exploration and utilization through a dynamic epsilon-greedy mechanism; and in combination with a vertex number priority scheduling strategy and a multi-dimensional quality award function, guiding an intelligent agent to adaptively generate a high-quality quadrilateral grid. According to the method, full-automatic and high-robustness subdivision of the complex polygon area is achieved, the grid orthogonality and the length-width ratio quality are remarkably improved, manual intervention is avoided, and the method is suitable for engineering scenes such as finite element analysis.

The invention discloses a start-stop channel decision-making method and system based on reinforcement learning, and the method comprises the steps: constructing a traffic flow interruption probability model through employing a Brilon model, and depicting a relation between a congestion probability and a traffic flow parameter; converting a start-stop channel decision problem into a Markov decision process, and determining a state space, an action space and a state transition probability of the Markov decision process; and solving the Markov decision process by adopting a Q-Learning algorithm to obtain a start-stop channel decision. According to the invention, the traffic condition in a period of time in the future can be accurately pre-judged, and the use strategy of the emergency lane is planned in advance; by optimizing the start-stop decision logic, unnecessary resource waste is reduced, meanwhile, the use reasonability and efficiency of the emergency lane are improved, and it is ensured that the lifeline is unobstructed at the key time.

The invention discloses an unmanned aerial vehicle trajectory optimization and power distribution method, system and device based on cost evaluation reinforcement learning, and a medium, and the method comprises the steps: constructing an unmanned aerial vehicle data service scene under safety constraints, modeling a multi-objective optimization problem of unmanned aerial vehicle trajectory optimization and power distribution, and designing an unmanned aerial vehicle data service evaluation index; converting a multi-objective optimization problem into a constrained Markov decision problem; respectively designing a state space, an action space, an instant reward function and a corresponding cost function; constructing a cost evaluation critic network parallel to the reward evaluation critic network, training and updating network parameters of the unmanned aerial vehicle by using a cost evaluation SAC algorithm according to a constrained Markov decision problem, and calculating trajectory optimization and power distribution parameters of the unmanned aerial vehicle; the system, the equipment and the medium are used for implementing the method. According to the invention, ground user data service requirements of a data service scene under random and dynamic security constraints are met.

The invention discloses an enterprise appeal intelligent perception and closed-loop processing system based on multi-modal AI, and relates to the technical field of intelligent government affairs, the system realizes accurate semantic understanding and deep intention recognition of multi-modal fusion, and the intelligent perception and closed-loop processing of enterprise appeals are realized through a cross-modal attention mechanism and a unified semantic representation model. According to the method, deep fusion and complementary analysis of multi-source heterogeneous data such as voices, texts, images and the like are realized, the limitation of a traditional single-mode or simple splicing mode is broken through, the accuracy and robustness of intention recognition are remarkably improved, and misjudgment is fundamentally reduced; the method comprises the following steps: constructing a dynamic self-adaptive routing mechanism based on Actor-Critic reinforcement learning, constructing a work order assignment problem into a sequence decision problem, driving a model to learn a dynamic fusion and weight distribution strategy of multiple decision factors through a reward mechanism, and adopting an online strategy iteration optimization mechanism to realize an optimal assignment decision, so as to improve the work order assignment efficiency. And the shunting accuracy and efficiency are obviously improved.

The invention provides a pre-disaster-post-disaster two-stage power distribution network investment decision-making method and system, and the method comprises the steps: generating a plurality of post-disaster fault simulation scenes and the occurrence probability of each post-disaster fault simulation scene for a worst typhoon event based on the analysis of historical typhoon data, and calculating the fault probability of a power distribution network line in the worst typhoon event scene; aiming at the fault condition, constructing an upper-layer optimization model by taking the minimum investment cost of pre-disaster element reinforcement and energy storage system configuration and the optimal preset toughness level as targets; and based on an optimal element reinforcement and energy storage system configuration planning scheme obtained by the upper-layer optimization model, constructing a lower-layer optimization model by taking the minimum loss load cost in different post-disaster fault scenes as a target, and solving the double-layer optimization model to obtain multiple pre-disaster investment decision-making schemes. According to the scheme, the influence of the post-disaster maintenance decision problem on pre-disaster investment planning is fully considered, and the pre-disaster investment cost and the post-disaster loss load cost are balanced based on the multi-objective optimization theory, so that multiple pre-disaster investment decision schemes are obtained, and the overall toughness level of the power distribution network is improved.

The invention relates to an average field game optimization method and system in an edge computing network based on content popularity. The method comprises the following steps: (1) setting a system model; comprising the steps that an edge computing network is set, and the edge computing network comprises a cloud data center, a plurality of edge nodes with storage units and a plurality of users; (2) modeling a video service-oriented cache decision problem in the edge computing network as a random differential game problem, namely an SDG problem, according to a residual storage space state of each edge node and a state change of a task popularity degree based on content popularity; and (3) solving the SDG problem by using a mean field game method, obtaining Nash equilibrium oriented to a video service cache decision problem in the edge computing network, and obtaining an optimal cache strategy of each edge node. According to the method, the satisfaction of user requirements is concerned, and the caching income of the edge node is considered.

The invention discloses an e-commerce platform product selection and pricing method based on multi-agent reinforcement learning, aiming at the game decision problem between an e-commerce platform and a supplier, a platform agent and a supplier agent are constructed, the platform agent uses a deep Q network DQN to learn a product selection strategy, and the supplier agent uses an Actor-Critic algorithm to learn a pricing strategy. Through Stackelberg master-slave game modeling, a platform is used as a leader to make a decision for selection, and a supplier is used as a follower to make a decision for pricing after observation. A state space (including commodity sales volume, inventory, user score, supplier fulfillment rate and the like), an action space (discrete commodity selection and continuous pricing) and a multi-target reward function (balance profit, user satisfaction, inventory turnover and the like) suitable for an e-commerce scene are designed. The strategies of the two parties are converged to game equilibrium through alternate training, and online continuous learning is supported to adapt to a dynamic market environment. Experiments show that compared with an existing method, the method has the advantages that the overall income can be improved by 18-30%, the win-win situation of the platform and the suppliers is achieved, and the blank of multi-agent reinforcement learning in e-commerce product selection and pricing scenes is filled.

The invention discloses an unmanned sweeping robot navigation method based on multi-sensor fusion, and belongs to the technical field of sweeping robots. According to the method, an environment sensing system is constructed through multi-source sensor fusion, and efficient path planning and obstacle avoidance decision making in a dynamic environment are realized in combination with a quantum heuristic behavior decision engine; a traditional decision problem is converted into quantum state superposition operation through a quantum state probability amplitude model, and dynamic coupling of environment perception and a task target is achieved through the joint effect of a complex phase angle and a weight coefficient; and the dynamic collapse equation improves the adaptive capacity of behavior selection. The sweeping efficiency, safety and environment adaptability of the sweeping robot in a complex dynamic environment are remarkably improved, and the sweeping robot is suitable for various scenes such as families and offices.

A multi-hop reasoning method based on dynamic reasoning guidance and multistage self-feedback retrieval belongs to the field of natural language processing, and comprises the following steps: deconstructing a multi-hop reasoning process into a target-oriented sequence decision problem, carrying out dynamic reasoning guidance by using a large language model, generating a sub-problem sequence matched with a reasoning progress in real time, and carrying out multi-level self-feedback retrieval on the sub-problem sequence; target document retrieval is guided, and sub-questions are dynamically generated; according to the generated sub-questions, obtaining associated documents by adopting a three-level collaborative retrieval mechanism; and performing information refining on the associated document through a large language model, fusing the refined information into an inference chain, and performing inference to generate an answer. The invention further discloses a multi-hop reasoning system, a storage medium and a computer program product. The method aims at solving the complex multi-hop problem that multiple dispersed knowledge fragments need to be integrated, high-accuracy and high-efficiency reasoning is achieved, the retrieval requirement is dynamically generated through an explicit thinking chain guiding mechanism, and evidence obtaining is optimized and redundant information is filtered in combination with a three-level self-feedback retrieval mechanism.

The invention provides a multi-agent task planning and executing method and a decision making system, and relates to the technical field of artificial intelligence and automatic control. According to the method, a complex multi-agent decision problem is decomposed into an upper-layer semantic planning subsystem and a lower-layer action control subsystem which are relatively independent and collaborative optimization; the upper-layer vision-language model planner generates sub-target sequences and track priori conforming to tactical intentions from a global perspective by utilizing the powerful vision understanding and language reasoning capabilities of the upper-layer vision-language model planner; and the reinforcement learning strategy controller at the lower layer focuses on executing specific control actions under the guidance of the sub-targets. Meanwhile, the system can optimize the final task performance and the process execution quality at the same time by introducing an evaluation mechanism and a combined reward signal based on the completion degree of the sub-targets. Environmental native rewards ensure that the system is optimized towards a final objective.

The invention belongs to the technical field of workshop production scheduling, and discloses a flexible job workshop multi-target scheduling method and system based on preference driving, and the method comprises the steps: defining a target function and a constraint condition based on obtained information, and constructing a workshop scheduling model; converting a workshop scheduling problem into a Markov decision problem, designing a reward function, and creating a preference pool; selecting a preference vector from the preference pool, and respectively inputting the preference vector and the system state into a strategy network and a value network for iterative training to obtain a trained strategy network and a trained value network; and obtaining current preferences for different dispatches and a current state of the system, inputting the current preferences and the current state of the system into the trained strategy network, obtaining action probability distribution in a given state, and selecting a dispatching action with the maximum probability to obtain a corresponding optimal dispatching scheme. According to the method, the requirements of tire enterprises in different situations are met, high flexibility is achieved, the nonlinear target utility function is designed, and a tire workshop scheduling scheme with higher quality can be found out easily.

An electric heavy truck collaborative scheduling optimization method based on multi-agent reinforcement learning belongs to the technical field of demand side resource scheduling and intelligent transportation, and comprises the following steps: firstly, modeling an electric heavy truck and a vehicle network interaction system, and depicting a constraint relationship among a transportation task, charging and discharging and power grid interaction; secondly, constructing a Markov decision process, defining a state, an action, a state transition rule and a reward function of a single agent, and converting a scheduling problem into a decision problem solvable by reinforcement learning; then, based on an MAPPO algorithm, a distributed execution-centralized judgment architecture is adopted, a multi-agent collaborative decision-making model is trained, decision-making feasibility is ensured through action masks, and collaborative optimization is achieved through global value evaluation; and finally, realizing dynamic scheduling and optimization decision of the electric heavy truck based on the training model. According to the invention, the transportation efficiency and the power grid interaction cost can be balanced while the transportation timeliness and the battery safety constraint are satisfied, and the overall operation economy and collaboration of the electric heavy truck fleet are improved.

The invention relates to the technical field of resource optimization decision, in particular to a multi-stage decision model construction method based on dynamic programming, which comprises the following steps: modeling based on resource constraint and an objective function, dividing a multi-stage decision problem into sub-problems, extracting logic association between states, identifying key nodes, and detecting path redundancy or missing. And evaluating resource allocation mode requirements, optimizing connection consistency between states, adjusting a decision model structure, identifying new mode attributes, and obtaining an updated multi-stage decision model. According to the method, the problems of key node identification and redundant path detection are effectively solved by accurately dividing the multi-stage decision problem and combining dynamic planning and analysis of the state transition path, the flexibility and accuracy of the decision process are improved, it is ensured that the decision is efficiently matched with external fluctuation all the time, decision lagging and low efficiency are avoided, resource allocation is continuously optimized, and the efficiency is improved. Coordination and consistency of all links are ensured, and the problems of resource waste and lag caused by the fact that real-time adjustment cannot be achieved in a traditional method are solved.

The invention relates to a low earth orbit satellite Internet of Things satellite-ground collaborative unloading method with optimized energy consumption. The method comprises the following steps: constructing an LEO satellite MEC network model and initializing system parameters; calculating a system energy consumption sum of user terminal UE calculation, idle terminal AE unloading calculation and LEO satellite MEC unloading calculation, and establishing an optimization problem; the whole problem is divided into five sub-problems, namely user matching, unloading decision, unloading proportion, and computing resource allocation of AE and LEO. Solving a user matching sub-problem by adopting a maximum weight matching algorithm; solving an unloading decision problem by adopting a relaxation and rounding technology; solving an AE and LEO satellite resource allocation scheme by adopting a Lagrange multiplier method and a KKT condition; and the optimal unloading decision, the optimal unloading proportion and the computing resource allocation of the AE and the LEO are obtained through alternate iterative optimization. The method has the remarkable advantages that the energy consumption problem existing in the satellite communication network can be effectively relieved, and the excellent performance is shown in the aspect of energy consumption reduction through the innovative technical means and the unique design thought.

The invention discloses an ultra-dense heterogeneous wireless network access control method fusing prediction and reinforcement learning, and the method comprises the steps: simulating a movement track of a user through employing an SLAW model, and carrying out the normalization of the movement track; inputting the normalized moving track into the LSTM model, and outputting a predicted coordinate of the next moment; calculating the received signal strength of the user to the current access network and all other candidate networks under the predicted coordinates, and judging whether to trigger a network switching mechanism in advance according to the received signal strength; for a network switching decision problem, constructing a Markov decision model by defining a state space, an action space and a reward function; when a network switching mechanism needs to be triggered in advance or a preset decision period is reached, solving the Markov decision model by using a DDQN algorithm, selecting a target network from the candidate networks, and switching to the target network; according to the invention, the response speed and accuracy of network switching can be effectively improved, so that the user experience and the network performance are improved.

The invention discloses an RSU auxiliary single-hop task unloading optimization method based on multi-agent reinforcement learning, and belongs to the technical field of Internet of Vehicles and edge computing. The method comprises the following steps: actively sensing a road environment through an RSU, and generating a sensing task; the distribution state of adjacent vehicles is predicted through a GRU model, and an adjacent node set of each RSU is updated in real time; according to task characteristics, tasks generated by the RSU are unloaded to adjacent edge computing nodes, and time delay and energy consumption unloaded to other nodes are calculated based on a time delay and energy consumption calculation model; optimizing the objective function to minimize the total time delay and the total energy consumption of calculation, and modeling a task decision problem as a Markov decision process; and performing distributed training and optimization by using an MAPPO algorithm to obtain an optimal task unloading decision. According to the method, the long-time average task processing time delay and energy consumption can be effectively reduced, and the task unloading efficiency and robustness in the dynamic Internet of Vehicles environment are improved.

The invention discloses a multi-target hierarchical depth optimization method for illumination control, and belongs to the technical field of illumination control, and the optimization method specifically comprises the following steps: I, collecting and preprocessing various types of information at edge nodes of an illumination area, and extracting semantic information describing a real-time scene from the preprocessed various types of information; according to the method, the influence of sensing noise, missing data and heterogeneous sampling on decision is remarkably reduced, the accuracy and stability of environment sensing are improved, the illumination control strategy has the immediate response capability and the prospective adjustment capability, the problem that a traditional fixed weight method is insufficient in adaptability in different time and different scenes is solved, and the method is suitable for popularization and application. According to the method, the targets of energy conservation, safety, comfort and the like can be dynamically balanced according to the actual environment, the complex decision problem is effectively decomposed, the convergence speed and robustness of the strategy are improved, meanwhile, sudden performance drop caused by scene extrapolation is avoided, and the generalization ability of the system in the complex urban environment is remarkably improved.

The invention discloses a rapid task planning method based on deep learning. The method comprises the following steps: firstly, constructing a generalized Voronoi graph on an occupied grid map obtained by exploration of multiple unmanned vehicles; secondly, quickly calculating the distance between each unmanned station point and each task point by using the topological grid to obtain a distance matrix; on the basis of the distance matrix, a task planning problem of the unmanned vehicle is modeled into a multi-objective optimization problem, the multi-objective optimization problem is divided into an upper-layer task allocation problem and a lower-layer path sequence decision problem, the upper layer adopts a multi-objective particle swarm optimization algorithm, tasks are reasonably allocated to appropriate unmanned vehicles, and the lower layer adopts a multi-objective particle swarm optimization algorithm. For each unmanned vehicle, a trained TSP deep learning solution model is adopted to calculate an optimal execution sequence; on the basis of obtaining a specific scheme, in a heuristic area of a topology network in an environment map, an RRT * algorithm is improved, and a planning specific execution path is obtained. According to the method, rapid path planning is realized, and high-quality cost is provided for task allocation.

The invention discloses a reinforcement learning-based household service knowledge graph dynamic updating method, and relates to the technical field of artificial intelligence and data processing. The method comprises the following steps: initializing a knowledge graph into a reinforcement learning environment; defining an agent state including a current entity, target information and a historical path; defining an action space of the intelligent agent; designing a reward function including a final reward and an intermediate reward; training an intelligent agent by adopting a strategy network; and inferring and verifying new knowledge by using the trained agent, and adding the new knowledge back to the knowledge graph. The knowledge graph updating process is modeled into a sequence decision problem, automatic discovery of knowledge and dynamic evolution of the graph are achieved through autonomous exploration and learning of the reinforcement learning agent, and the problems that a traditional updating method is high in cost and poor in timeliness are effectively solved.

The present application discloses an optimization control method, device and medium for unmanned aerial vehicle (UAV) information collection, which includes: defining the information collection method of the UAV, establishing a set of UAV flight actions, and constructing an optimization control problem for UAV information collection; obtaining the action information and location information taken by the UAV at a certain moment, and constructing a model for the total amount of information collected by the UAV; constructing a sequential optimization decision model based on the model for the total amount of information collected by the UAV, and modeling the UAV information collection optimization control problem as a problem of finding the optimal strategy for the sequential optimization decision problem; and reconstructing the sequential decision problem as a steady-state reinforcement learning problem and establishing a robust reinforcement learning model; based on the robust reinforcement learning model, using a robust Q-value learning algorithm to solve the problem, and obtaining the optimal control strategy for information collection, which can collect information quickly and effectively, adapt to various scenarios and environments, and have good scalability.

The invention provides an intelligent decision-making control method and system based on the motion trail of a mechanical arm, and belongs to the technical field of intelligent. The method comprises the steps that a first method and a second method are executed at the same time; comprising the steps that according to motion track data of a second mechanical arm, an improved dynamic time warping algorithm is adopted, and the similarity between a first real-time track and a standard action curve is obtained; the motion track data of the second mechanical arm is converted into a sequence decision problem, and the similarity between a second real-time track and a standard action curve is obtained according to the value of a reward function in the sequence decision problem; and under the condition that the similarity is smaller than or equal to a preset similarity threshold value, a third method is adopted for dynamically correcting the motion track data of the mechanical arm. According to the method, on the premise that the real-time performance is met, the conformity degree of the actual task execution condition of the model and the expectation can be measured more accurately.

The application belongs to the technical field of natural language processing and knowledge graph, and discloses a kind of multi-modal acquisition and evolution graph analysis method and system of policy text. In view of the passive decision problem caused by the rigidity of existing policy analysis, semantic understanding and lack of dynamic correlation, the application adopts visual and DOM feature weighted fusion and reinforcement learning agent to realize multi-modal adaptive acquisition and joint alignment in anti-crawling environment;Use large language model for semantic deconstruction, and convert implicit logic into computer-readable explicit logic expression through thought chain reasoning;Construct a policy evolution knowledge graph with time and validity level attributes, perform satisfiability analysis to output conflict detection results;Map the target object portrait to the graph and the logic expression to evaluate item by item, and output intelligent decision deduction results combined with quantitative dimensions. The application realizes the whole life cycle intelligent analysis of policy data acquisition, deep logic analysis and active deduction.

An unmanned aerial vehicle resource allocation method, system and device based on multi-agent reinforcement learning, and a medium, the method comprising: establishing an optimization problem of multi-unmanned aerial vehicle resource allocation for a multi-modal unmanned aerial vehicle assisted communication-sensing integrated data service scene, and designing an unmanned aerial vehicle data service evaluation index of average peak information age; expressing an optimization problem as a Markov decision problem, and designing a state space, an action space and a corresponding instant reward function; the method comprises the following steps: designing a conditional variation auto-encoder, representing a coupling relationship between discrete actions and continuous actions, combining an MATD3 multi-agent reinforcement learning algorithm to obtain an MATD3 algorithm represented by the coupling relationship, and training and updating network parameters of an unmanned aerial vehicle by using the MATD3 algorithm represented by the coupling relationship according to a Markov decision model. The method is applied to a data service scene for communication resource allocation; according to the method, the data service requirements of the ground users in a random and dynamic communication and sensing integrated scene can be better met.