131 results about "Distributed decision" patented technology

The invention relates to the technical field of robot scheduling, in particular to a distributed robot collaborative scheduling system and method in a dynamic environment, and the system comprises an environment sensing layer which is used for collecting environment dynamic data in real time; the distributed decision-making layer comprises local task scheduling modules of a plurality of robots; the cooperative communication layer is used for realizing task state synchronization and conflict detection among the robots based on a low-delay communication protocol; the dynamic weight calculation module is used for generating a real-time optimization weight according to the task emergency degree, the robot energy consumption and the path risk factor; according to the invention, by using a completely distributed collaborative scheduling architecture, through a decentralized task distribution mechanism and a distributed consensus protocol, a single-point fault risk existing in a traditional centralized scheduling system is thoroughly eliminated, and even if a part of robot nodes have faults or communication is interrupted, the system can work normally. And the system can still run continuously through autonomous negotiation of the remaining nodes, so that the reliability of the system in a complex environment is remarkably improved.

The invention discloses a multi-aircraft distributed decision-making swarm route avoidance cooperative control method and system, relates to the technical field of unmanned aerial vehicle formation control, and is used for multi-aircraft flight formation. Each unmanned aerial vehicle is provided with an intelligent decision-making module with environment sensing, communication and path planning capabilities; the method comprises the steps that each unmanned aerial vehicle generates a preliminary avoidance path by adopting an improved genetic algorithm based on acquired local environment information and received state information of neighborhood unmanned aerial vehicles in combination with a preset avoidance rule and a preset multi-objective optimization function; the optimization objectives of the preset multi-objective optimization function comprise an obstacle avoidance safety distance, an energy consumption coefficient, a formation retention degree and task timeliness; space-time conflict detection is executed based on the multiple preliminary avoidance paths, the preliminary avoidance paths are optimized based on conflict detection results, a final conflict-free optimal path is obtained and broadcasted to other unmanned aerial vehicles in the formation, and distributed collaborative decision making is achieved; according to the invention, the control efficiency of multi-unmanned aerial vehicle formation control is improved.

The invention provides a production line process scheduling optimization method and system based on an agent cluster. The method comprises the steps that the agent cluster comprising a plurality of distributed decision agent nodes is configured according to a production line process scheduling instruction; based on the production line architecture feature graph, a communication protocol between agent nodes is generated through node association relationship analysis, and a cooperation rule set is constructed in combination with a process priority rule and a resource constraint condition; the intelligent agent cluster is instructed to perform scheduling according to a communication protocol and a cooperation rule set, process tasks are dynamically allocated and adjusted through real-time state interaction among nodes, scheduling behavior logs are obtained, a reinforcement learning training sample set is constructed in combination with expected labels, decision parameters of intelligent agent nodes are optimized by using a multi-intelligent-agent reinforcement learning algorithm, and a reinforcement learning training result is obtained. According to the method, the dynamic response capability and the cooperation efficiency of automatic production line scheduling can be improved, and the method is suitable for efficient process scheduling in a complex production environment.

The invention relates to the technical field of unmanned aerial vehicles, in particular to an intelligent dynamic pilot frequency networking system and method for an image transmission module MESH ad hoc network. The method is based on spectrum fingerprint features, combines a real-time interference topological graph constructed by interaction of neighborhood nodes, identifies a conflict frequency set needing to be evaded through multi-dimensional weighted evaluation, and dynamically generates a pilot frequency strategy table containing main and standby frequencies and a switching threshold value by using a frequency hopping decision engine according to the conflict frequency set and a preset legal frequency pool. Synchronizing the pilot frequency strategy table to an associated relay node, generating a link stability evaluation matrix and feeding back the link stability evaluation matrix to a frequency hopping decision engine; and correcting a frequency switching threshold value and an alternative frequency weight in the pilot frequency strategy table based on the link stability evaluation matrix. According to the invention, the spectrum utilization efficiency of the unmanned aerial vehicle image transmission system can be improved, the stability and reliability of a multi-hop link are ensured, the networking time delay is reduced through a distributed decision-making mechanism, and an effective solution is provided for high-quality wireless transmission in a dynamic topology environment.

The invention relates to the field of multi-source data management, in particular to an intelligent security collaborative management system based on multi-source perception and a large language model. Comprising a multi-source data acquisition module which is accessed to various intelligent monitoring devices and is output and converted into a unified standard tuple format through a mapping function; the information fusion module is used for screening a candidate alarm set according to the space-time tolerance, and generating composite alarm information by adopting confidence ranking and semantic embedding weighted averaging; the RAG knowledge base module is used for generating a composite event description through a large language model and vectorizing the composite event description as a retrieval index; the intelligent retrieval module is used for acquiring related historical events by adopting a mixed retrieval strategy and constructing structured cue words; the LLM decision module is used for outputting root cause analysis and classification disposal suggestions based on the composite alarm and the priori knowledge; the double-path response module is used for distributing decision suggestions to management personnel and an agent system to realize collaborative execution; and the feedback optimization module is used for collecting disposal data and adjusting the weight of the knowledge base and the decision template to realize continuous optimization.

The invention discloses a distributed collaborative decision-making system based on multi-modal data driving and an implementation method thereof, and relates to the technical field of group intelligence and distributed decision-making, and the system comprises a user end interaction module which provides a multi-modal interaction and decision-making scheme visual interface; the distributed node management module comprises a main node and an edge node, and the main node manages node registration, state monitoring and task distribution; the information fusion and preprocessing module is used for processing multi-source heterogeneous data; the decision analysis module is used for carrying out clustering analysis on the opinions and generating candidate schemes in combination with domain knowledge; the domain knowledge graph module is used for constructing a domain entity relationship network; the consensus mechanism and credit evaluation module determines multiple rounds of interaction rules, calculates a user credit value and influences an opinion weight; and the decision result output and feedback module is used for collecting user feedback for system optimization. According to the method, the stability, the response speed and the load balancing capacity are improved, the multi-source information processing and opinion aggregation quality is optimized, and efficient and reliable support is provided for distributed collaborative decision making.

The invention provides an active power distribution network multi-time scale rolling optimization method driven by multi-agent graph reinforcement learning. The method comprises the following steps: dividing slow time scale equipment and fast time scale equipment; the slow time scale is optimized and modeled into a multi-agent Markov game, and a global state space and a combined action space corresponding to switch actions, reactive compensation equipment gears and voltage regulation equipment gears are defined; switching action candidate vectors are generated based on a basic loop of the power distribution network, and action space dimensions are compressed through a scheme of excluding conflicting actions of the same common branch; fitting an action value function of each agent by adopting a graph neural network, generating a dimensionality-reduced discrete action strategy according to the global state space, and carrying out collaborative optimization through a distributed decision and a global reward mechanism; and on the basis of the slow time scale optimization result, the fast time scale optimization problem is converted into a mixed integer second-order cone programming model, and an instruction is solved in a rolling manner.

The invention discloses a source network load storage real-time scheduling method for MADRL-GAN collaborative optimization, and the method comprises the steps: firstly constructing a carbon pollution collaborative optimization model, and converting the model into a solvable convex problem through function linearization and mixed integer conversion; further mapping the model into a multi-agent reinforcement learning model, dividing agents according to an electrical coupling degree, and designing a state and action space containing a dual-Critic reward mechanism; thirdly, learning system uncertainty distribution by using a generative adversarial network, generating diversified scenes to train a reinforcement learning model, and obtaining a preliminary strategy; and finally, correcting the strategy through the generative adversarial network, generating a final scheduling action through a strategy mixing mechanism, and realizing distributed real-time optimal scheduling. According to the method, the problems of multi-target collaboration, high-uncertainty scheduling and real-time distributed decision making in a high-proportion renewable energy system are effectively solved.

The invention provides an intelligent management system and method based on multi-agent cooperation. The intelligent management system comprises a data acquisition module used for acquiring production field data in real time; the data processing module is used for data cleaning; the large model module is used for performing semantic analysis on a task input by a user based on a large model; the agent module is used for performing task scheduling based on a TSQlearning scheduling algorithm; and the RAG knowledge base module is used for constructing an RAG knowledge base and providing knowledge retrieval based on hierarchical indexing and a mixed retrieval mechanism. According to the method, user requirements are analyzed through the large model module, the task comprehensive scheduling capability is improved through the distributed decision of the agent module and the TSQlearning algorithm, the resource utilization rate is improved, the information retrieval accuracy is improved from 70% to 92% through the hierarchical indexing and mixed retrieval mechanism of the RAG knowledge base, and the management efficiency and the intelligent level of the smart factory are improved.

The invention relates to the technical field of unmanned aerial vehicle cluster energy management, and discloses an unmanned aerial vehicle ad hoc network-based autonomous inspection and charging nest system, which comprises the following steps that: each unmanned aerial vehicle in a cluster calculates and broadcasts a time dynamic scalar capable of being freely controlled in real time, and broadcasts a task and an energy opportunity packet when the scalar of a certain unmanned aerial vehicle is lower than a threshold value; according to the invention, through a distributed decision-making mechanism driven by an energy potential difference, cluster energy is enabled to dynamically flow according to a minimum resistance path like liquid, internal energy consumption caused by a fixed return path in traditional inspection is avoided, and the reliability of the unmanned aerial vehicle is improved. And meanwhile, by combining dual-channel communication arbitration and a mechanical locking unit, reliable energy cooperative transmission can still be maintained in a complex environment.

The invention discloses an artificial intelligence content generation task multi-server collaborative optimization method based on deep reinforcement learning, and relates to the crossing field of edge computing and artificial intelligence. According to the method, a user equipment-base station-edge server-core network four-layer architecture is constructed, distributed decision is realized through a deep Q network model, and the method comprises the following steps: defining a task and time delay model; a deep reinforcement learning framework is designed, a decision model is constructed through a state space, an action space and a reward function, and training stability is improved by adopting experience playback and a target network slow update mechanism; and providing an adaptive multi-server selection and load distribution strategy, and solving an optimal distribution proportion based on a time delay equality principle. According to the method, the average unloading time delay of the AIGC tasks is remarkably reduced, the task failure rate is reduced, the dynamic environment adaptability and the extreme scene robustness are improved, and the method is suitable for scenes with strict requirements for time delay and reliability.

The invention provides a traffic corridor signal cooperative control method based on single agent reinforcement learning, and relates to the technical field of traffic management and control, and the control method comprises the steps: obtaining the real-time traffic state data of a traffic corridor comprising a plurality of signal intersections; acquiring a current signal control scheme of the traffic corridor; constructing a state vector according to the real-time traffic state data and the current signal control scheme; inputting the state vector into a pre-trained single-agent reinforcement learning model to obtain a corresponding action vector; based on the action vectors, phase division of all the signal intersections is synchronously adjusted, and a new signal control scheme is generated; according to the invention, signal timing of all signal intersections in a traffic corridor is cooperatively controlled by adopting a centralized single-agent architecture, so that the problems of system complexity and training instability caused by local observation, distributed decision and communication coordination among agents in a multi-agent scheme are fundamentally avoided.

The invention belongs to the technical field of collaborative unmanned aerial vehicle cluster reinforcement learning, and relates to an unmanned aerial vehicle cluster dynamic strategy optimization method based on a super network and a diffusion model. A dynamic weight generator (super network) is embedded in a diffusion strategy model (MADiff) framework, a differentiated Actor network weight is generated for each agent, and strategy personalized customization is realized, so that the problems of cooperation failure and training oscillation caused by strategy homogenization in a traditional method are solved. The dynamic strategy optimization normal form is especially suitable for continuous operation of a large-scale unmanned aerial vehicle cluster in a non-stationary environment, and a reinforcement learning framework with environment perception-strategy prediction-dynamic adjustment closed loop is constructed. The performance of the system in the aspects of dynamic task collaboration, non-stationary strategy optimization, distributed decision consistency and the like is remarkably improved, and key technical support is provided for application of an intelligent unmanned cluster in a complex scene.

The invention discloses an unmanned aerial vehicle autonomous cluster cooperative control system based on swarm intelligence, and belongs to the technical field of unmanned aerial vehicle control. The control system comprises an unmanned aerial vehicle cluster which is composed of a plurality of unmanned aerial vehicles with autonomous flight capability; each unmanned aerial vehicle is provided with a sensing unit, a navigation and flight control unit, a communication unit and a cluster cooperative control unit; a group intelligent decision module based on an ant algorithm is arranged in the cluster cooperative control unit and comprises a virtual pheromone generation and updating sub-module, a local decision sub-module and a task planning and cooperative sub-module; the system provided by the invention takes swarm intelligence as a core, realizes distributed decision-making through a virtual pheromone field, avoids communication and computing power bottlenecks of centralized control, and also can enable a cluster to autonomously emerge a global optimization behavior. Through dynamic role allocation, task decomposition and priority scheduling mechanisms, resources can be adaptively allocated along with task progress; and the collaborative behavior pheromones strengthen the efficient path, so that the execution efficiency of the cluster in the complex task is remarkably improved.

The invention relates to the technical field of fluid machinery control, and discloses an intelligent vibration reduction control method and system for a centrifugal pump, and the method comprises the following steps: obtaining vibration signal data to obtain frequency spectrum characteristics; when the spectrum feature exceeds a threshold value, abrupt change parameters are extracted, pipeline system coupling data are fused, and a multi-pump cooperation state vector is generated; generating a distributed decision instruction sequence according to the state vector classification; fusing the feedback information according to the instruction sequence to obtain an initial adaptive strategy parameter; the operation parameters are adjusted according to the initial parameters, the efficiency is calculated, loop iteration optimization is carried out until the efficiency reaches the standard, and final self-adaptive strategy parameters are obtained; and the stability is verified after the final parameters are deployed, and loop optimization is returned if the verification fails. Through multi-pump cooperation and double closed-loop optimization, the stability, efficiency and reliability of cooperative operation of the pump set can be improved.

The invention discloses a multi-robot control system, and relates to the field of robot path planning. According to the system, in a robot path driving task, candidate path nodes are selected by randomly switching target guidance and a global exploration mode, feasible nodes and path segments are determined by combining historical path nodes, and the path nodes are expanded after collision detection of a static obstacle and a dynamic robot. The system also dynamically adjusts a driving step length based on a load rate, sets a safety margin including load compensation, and realizes multi-machine cooperation through local path topological optimization and distributed decision. According to the method, the multi-robot path conflict and collision risk is effectively reduced, the heavy-load working condition motion stability and the system cooperation efficiency are improved, and the method is suitable for a multi-robot cooperation operation scene.

The invention relates to a workshop scheduling system and method based on combination of a graph neural network and multiple agents. The system mainly comprises three modules: a graph neural network module, a multi-agent system module and a scheduling decision and optimization module. The graph neural network module extracts features of a complex relationship between a task and a machine through a double attention network (DAN). And the multi-agent system module performs distributed decision making and scheduling optimization based on a reinforcement learning strategy by using the feature information. And the scheduling decision and optimization module dynamically adjusts and optimizes the scheduling decision according to the real-time production condition, so that the efficient operation of the production process and the optimal utilization of resources are ensured. The method aims at improving the scheduling efficiency and optimizing the scheduling effect, remarkably improving the production efficiency of a workshop and reducing the production cost.

The invention relates to a cluster cooperation-oriented unmanned aerial vehicle dynamic task allocation and path planning method and device, and the method comprises the steps: modeling a monitoring region comprising an unmanned aerial vehicle cluster, a plurality of to-be-tracked targets and a specified moving target as a partially observable Markov decision process, and enabling each unmanned aerial vehicle to obtain local observation information based on relative parameters; then, each unmanned aerial vehicle outputs a flight orientation adjustment instruction through a strategy neural network, and the network is trained and optimized under the guidance of a value network through a multi-agent near-end strategy optimization algorithm; the unmanned aerial vehicle cluster executes a joint action and generates a comprehensive reward signal fusing global task rewards, individual behavior rewards and anti-collision punishment; and finally, driving the strategy network to synchronously complete implicit task allocation and collaborative path planning through continuous feedback of the reward signal. According to the invention, integrated cooperative solution of task allocation and path planning is realized, so that the unmanned aerial vehicle cluster has adaptive cooperative capability and high robustness under a distributed decision framework.

The invention discloses a three-dimensional space unmanned aerial vehicle path planning method based on a swarm intelligence algorithm. The method comprises the following steps: S1, defining an initial boundary and sampling in the initial boundary to generate an initial population containing a plurality of candidate paths; s2, constructing a target function according to the optimization target; fitness values of the candidate paths are calculated based on the target function and the position function, and the candidate path with the high fitness value is selected as a global optimal solution; s3, the maximum number of iterations, the maximum number of iterations and the number of iterations of dynamic updating are set, and the specific number of iterations and the target probability are preset; introducing two dynamic parameters and designing a boundary adjustment coefficient, an adaptability factor and a dynamic probability which change along with the current iteration number; according to the method, autonomous navigation and path optimization of the unmanned aerial vehicle in a complex environment are realized by simulating the self-organization and distributed decision-making mechanism of group behaviors in nature, the path planning calculation speed can be effectively improved, the distance is reduced, and the adaptive capacity to environmental changes is enhanced.

The invention discloses an unmanned aerial vehicle dynamic path planning method and system based on artificial intelligence, and the method comprises the steps: collecting environment data in real time through a multispectral sensor and a laser radar carried by an unmanned aerial vehicle, and constructing a high-precision three-dimensional terrain grid map; s2, based on a deep learning image segmentation network, performing semantic segmentation on dynamic obstacles in the real-time video stream, and identifying moving vehicles, flying birds and sudden risk sources; s3, establishing a sudden risk source threat quantitative model: inputting a semantic segmentation result in the step S2 into a risk assessment LSTM network, and generating a real-time threat coefficient matrix of a dynamic obstacle in combination with wind speed and visibility meteorological data; the distributed decision is realized through a federated edge computing architecture, so that the path updating frequency is improved to 5Hz, the formation control delay is greatly reduced, and the out-of-control rate compression of a complex urban scene is relatively low.

The invention belongs to the field of intelligent traffic, and particularly relates to a construction area traffic flow guiding and speed cooperative control system which realizes accurate traffic flow control through cooperative operation of an area sensing module, a distributed guiding module, a communication module and an emergency response module. The area sensing module is based on three-dimensional modeling and digital twinning technologies, fuses real-time traffic flow, vehicle states and environment data, and constructs a three-dimensional simulation space of a target area; the distributed guide module establishes a drainage control model by using a distributed reinforcement learning algorithm, dynamically optimizes a sub-region congestion control strategy in combination with a vehicle-road-cloud three-level communication node, and adjusts the acceleration and speed of a vehicle in real time; and the emergency response module rapidly generates an emergency instruction through an anomaly analysis algorithm and a strategy library and feeds back the emergency instruction to the guide model to solve an emergency abnormal event, and the application integrates multi-source data real-time sensing, distributed decision optimization and a multi-stage communication architecture, and realizes dynamic traffic flow distribution and intelligent traffic control in a construction area.

The invention belongs to the technical field of cloud side-end collaborative reasoning, and discloses a cloud side-end model reasoning joint optimization method under air-ground collaboration. And designing a network architecture modeling module, an inference performance modeling module, a delay modeling module and a joint optimization module. An air-ground cooperative reasoning system is constructed, a thinking chain prompt mechanism is introduced to perform modeling on reasoning accuracy, and unmanned aerial vehicle selection, language model selection, reasoning task unloading decision and unmanned aerial vehicle trajectory are jointly optimized to minimize the total cost of the system. A continuous convex approximation method is adopted to optimize the trajectory of the unmanned aerial vehicle, and a multi-agent reinforcement learning method is combined to carry out distributed decision making and centralized training, so that low-delay and high-precision collaborative reasoning service is realized. According to the method provided by the invention, communication, calculation and resource reasoning are effectively coordinated in a cloud edge-end coordination scene with dynamic change of user requests and various task types, the overall service quality and resource utilization efficiency of the system are remarkably improved, and the method is superior to other existing methods.

The invention relates to a virtual power plant-oriented data center resource scheduling method, equipment and medium, and the method comprises the steps: carrying out the multi-time scale division and dynamic characteristic decoupling of IT equipment, energy storage system and refrigeration system resources, and forming a decoupling resource model; dividing a high-collaboration resource cluster, and calibrating an elastic capacity boundary of the resource cluster; solving a multi-market collaborative optimization model in a global optimization layer at a first preset time scale according to the electric power information, the carbon information and the computing power demand information, and generating a day-ahead baseline strategy; carrying out distributed decision making by adopting a federal reinforcement learning framework at a local decision making layer based on the day-ahead baseline strategy and the real-time operation data at a second preset time scale, and generating a real-time regulation and control instruction; and when a trigger event is monitored, triggering the global optimization layer to re-plan the day-ahead baseline strategy, issuing a re-planning result to a local decision-making layer, and updating a real-time regulation and control instruction. Compared with the prior art, the method has the advantages of high reliability, collaboration, robustness and the like.

The invention discloses an intelligent warehouse logistics real-time goods allocation optimization method and system, relates to the technical field of intelligent warehouse logistics, and discloses the intelligent warehouse logistics real-time goods allocation optimization method and system. By obtaining real-time warehousing data, constructing a warehousing dynamic state map, generating an initial allocation scheme, inputting the initial allocation scheme into a distributed decision-making network to carry out parallel collaborative evaluation optimization and issuing a global instruction, the problems of delay and conflict in peak hours of a traditional method are solved, and the warehousing operation efficiency and the system throughput can be improved.

The invention relates to the technical field of wireless communication, and discloses an intelligent traffic scheduling method and system based on an attention mechanism and deep reinforcement learning, and the method comprises the steps: capturing the fine-grained traffic features through a local attention mechanism, carrying out the deep understanding of a global attention mechanism on a network overall traffic mode, and carrying out the real-time traffic scheduling of the network. Local details and global context information are integrated and combined to extract flow characteristics of the network, a central coordination-distributed decision-making system of multiple collaborative optimization units is adopted, the flow characteristics are used as input, and strategy actions obtained by each collaborative optimization unit through a strategy network are subjected to strategy optimization. The value network fits a global state value function to evaluate the current action and environment state and feed back the current action and environment state to the strategy network, a random gradient strategy is adopted to update a strategy, and a scheduling decision is continuously optimized. Through multiple times of training and iteration of central coordination-distributed decision making, optimization and coordination of a collaborative optimization unit are realized, the network state is continuously monitored, and a real-time decision is made for analyzing the network flow.

The invention provides an unmanned aerial vehicle cluster distributed region search method and system based on role differentiation. The method comprises the following steps: initializing an unmanned aerial vehicle collaborative search task environment; the information unmanned aerial vehicle updates a cluster search cognitive map; task allocation modeling of unmanned aerial vehicle cluster collaborative search; unmanned aerial vehicle tasks are pre-allocated based on CBAA information; updating a cluster topology connection state of the information subgroup; relay unmanned aerial vehicle task allocation based on connectivity maintenance; unmanned aerial vehicle waypoint tracking control based on collision time cooperative guidance; adjusting an information-relay subgroup proportion based on a region coverage rate; and outputting an unmanned aerial vehicle cluster collaborative search result. According to the method, the CBAA algorithm supports the communication of the unmanned aerial vehicle cluster to keep a distributed decision through improved strategies such as removal of disconnected individuals, an action selection tabu table and mobile base station switching; meanwhile, heterogeneous cooperation between information-relay subgroups is utilized, the action space of the unmanned aerial vehicle is released, the persistent exploration capability of a later cluster is enhanced, and local optimum of a search process is avoided.

The invention provides an intelligent terminal outdoor cooperation method and system based on an ant colony algorithm. The method comprises the following steps: acquiring outdoor navigation data, processing the outdoor navigation data into navigation data fragments, and distributing the navigation data fragments to a plurality of intelligent terminals; screening out a main relay node; when entering the target area, determining the intelligent terminal as main navigation equipment, and controlling the gas intelligent terminal to enter a low-power-consumption mode; summarizing the navigation data fragments stored in the intelligent terminals through the main relay node, forwarding the summarized navigation data fragments to the main navigation equipment, and splicing to generate complete navigation data of the target area; and the main navigation equipment obtains and generates navigation information and synchronizes the navigation information to each intelligent terminal through the main relay node. Through pheromone dynamic iteration of the ant colony algorithm and a distributed decision-making mechanism, efficient cooperation of multiple intelligent terminals on division storage and group collaborative splicing is realized, and on the premise of guaranteeing complete navigation data splicing precision and real-time navigation reliability, the overall energy consumption is remarkably reduced, and the equipment endurance is prolonged.

A distributed control method for urban multi-intersection traffic lights based on an evolutionary Q-network includes the following steps: constructing a Markov model for coordinated signal control, defining the agent, environment, state, behavior, and reward to provide a basic model for traffic light control; using a state encoding method to convert the actual traffic state into a binary feature vector, providing standardized input for the Q-network; based on the feature extraction results, using a Q-network structure encoding method to convert the neural network structure parameters into binary codes, preparing for the automatic search for the optimal network structure; using a Q-network structure evolution algorithm to automatically adjust the Q-network structure based on the state codes to find the optimal solution, automatically matching the Q-network structure with the complexity of the coordinated signal light control task; and constructing a distributed decision-making and computational method based on the road network subunits controlled by the Q-network to achieve coordinated signal light control on a large-scale road network. This method can improve road network load balancing and reduce transportation delays.

The application provides an intelligent unmanned system decision reliability evaluation and optimization method based on reinforcement learning, and belongs to the technical field of intelligent unmanned systems. The application adopts a multi-layer operation network model based on a coloring graph theory to represent individual autonomy and functional diversity, overcomes the limitation of a single-layer network, and provides a basis for decision analysis. An intelligent unmanned system distributed decision model based on an enhanced actor-critic architecture is adopted to improve the strategy performance and learning stability of individual decision. An intelligent unmanned system decision reliability evaluation model oriented to an operation loop is adopted. The model integrates the proposed reliability indexes into the individual learning process to improve the effectiveness of evaluation. The model successfully quantifies the influence of individual decision on system reliability. A decision reliability optimization model based on a cooperative multi-agent deep deterministic policy gradient algorithm is adopted. The proposed cooperative reward is integrated into the policy gradient optimization process to optimize the decision reliability.

The invention provides a sub-system cooperative control method, and belongs to the technical field of intelligent control, and the method comprises the steps: obtaining the regulation and control parameters of a target space where a sub-system is located; performing space-time alignment and feature fusion on the regulation and control parameters to generate a dynamic feature vector containing interaction influence factors of each subsystem; performing multi-objective optimization on the dynamic feature vector based on a deep reinforcement learning model to generate a collaborative strategy set including subsystem control instructions, execution priorities and weight allocation; and decomposing the collaborative strategy set into a subsystem control instruction sequence through a distributed decision framework, and performing dynamic feedback adjustment. According to the invention, the energy consumption of each subsystem can be reduced while multiple competitive indexes are synchronously and dynamically balanced; the problem of data asynchronization caused by network delay is effectively solved, and the precision of a control strategy is remarkably improved; and the improvement of an intelligent scene intelligent control technology is facilitated.