51 results about "Collaborative intelligence" patented technology

The invention relates to a network flow anomaly detection strategy generation method based on machine learning, and belongs to the technical field of machine learning. The method comprises the following steps: firstly, collecting network traffic data in a preset time window, and extracting feature vectors containing traffic, a time sequence and a protocol type; and inputting the feature vector into a long short-term memory auto-encoder model, and calculating a reconstruction error to judge whether the network flow is abnormal or not. Aiming at the abnormal feature vector, adopting a multi-agent depth deterministic strategy gradient algorithm to construct a plurality of cooperative agents, and independently generating a candidate abnormal detection strategy by each agent; through a cross-agent strategy evaluation mechanism, the difference between a joint strategy and a single-agent strategy in the aspect of anomaly detection accuracy is compared, cooperation gain is calculated, strategy exploration parameters of all agents are adjusted according to the cooperation gain, and a global optimal anomaly detection strategy is optimized and determined in real time. According to the method, high-precision and low-missing-report network traffic anomaly detection can be realized, and the method has good self-adaptability and real-time performance.

The invention relates to an Internet of Vehicles multi-agent cooperative sensing method and system. The method comprises the following steps: constructing a collaborative sensing network, wherein the collaborative sensing network comprises a self-agent and a plurality of collaborative agents; acquiring and processing sensing data through the collaborative sensing network; performing feature extraction to obtain intermediate features; self-adaptive sparsification is carried out to obtain sparse features, and the sparse features are compressed and transmitted to a self-agent; performing time sequence feature enhancement on the features of all the agents at the self-agent end; fusing the features to obtain fused features; and constructing an Internet of Vehicles perception model, and realizing perception by the detection model according to the fused features. According to the method, the calculation complexity of traditional global attention is reduced from the square level to the linear level through an adaptive sparsification mechanism, the calculation overhead is remarkably reduced while the multi-agent feature interaction precision is kept, and the method is more suitable for real-time operation on the vehicle-mounted edge equipment with limited resources.

The invention discloses an AI-based browser user behavior recording intelligent workflow method and system, and relates to the technical field of workflow management.The method comprises the steps that interaction behaviors are recorded in a browser, and normalized event data are generated; the input recording cooperative agent performs user task intention analysis and abstracts the user task intention into a structured workflow; inputting a recording test agent for verification, and adjusting the structured workflow to obtain a verified target workflow; the browser is driven to automatically execute and monitor the execution state, a self-repairing strategy is triggered for adjustment when the execution state is abnormal, and the target workflow is updated when self-repairing succeeds. Through the technical scheme of the invention, standardized modeling and precipitation of the browser event flow are realized, the workload of establishing an automatic process is reduced, the use threshold of establishing the browser automatic process is reduced, the execution success rate and stability of the target workflow are improved, and the user experience is improved. And the robustness and the long-term maintainability of the automatic process are effectively improved.

The invention discloses an enterprise financial risk determination method and a related device, and relates to the technical field of artificial intelligence. Six dimensions of mobility, credit, market, operation, industry macroscopic and public opinion reputation are respectively processed through six special and precise agents including a mobility risk analysis agent, a credit risk analysis agent, a market risk analysis agent, an operation risk agent, an industry macroscopic risk analysis agent and a public opinion and reputation risk analysis agent; and full-quadrant scanning of enterprise financial risks is realized. The cross validation coordinator and the dynamic cooperation engine form a double-wheel drive, and the cross validation coordinator can obtain a cooperation agent identifier set, a cooperation task type and cooperation configuration information; the dynamic collaboration engine dispatches multi-agent parallel collaboration in real time based on the collaboration agent identifier set, the collaboration task type and the collaboration configuration information, information islands are eliminated, and misjudgment and missing report are remarkably reduced. Therefore, accurate enterprise financial risks can be obtained.

The invention relates to a heterogeneous cluster cooperative task allocation and path planning method for an agricultural unmanned aerial vehicle, and the method comprises the following steps: S1, carrying out the system modeling and initialization, and carrying out the modeling of a cooperative data collection task of an unmanned aerial vehicle mechanism cluster in an agricultural environment as a decentralized partially observable Markov decision process; s2, defining a multi-dimensional space, wherein the decentralized part observes a global state space, a joint action space and a local observation space of a Markov decision process; s3, designing a mixed reward function; s4, performing cluster strategy training; and S5, performing online distributed execution. The method effectively solves the problems that an existing method is poor in environment dynamic change adaptability, does not fully consider the heterogeneous characteristics of the unmanned aerial vehicles and does not fully consider decision-making and execution separation, and remarkably improves the operation efficiency, safety and collaborative intelligence level of an agricultural unmanned aerial vehicle cluster in a complex unstructured environment.

The invention discloses a robot-assisted operation error detection method based on cooperative agent reasoning. The method comprises the following steps: constructing a multi-classification operation error system and a clinical informed knowledge base; based on a multi-classification operation error system and a knowledge base, a chain type reasoning prompt base is constructed according to different professional levels and analysis perspectives; for an input operation video clip, by calculating a composite risk score of a technical complexity score and a clinical influence score, allocating error instances of different risk scores to corresponding professional level reasoning paths; and deploying a time analysis agent, a space analysis agent and a process analysis agent for each professional level reasoning path, calling a corresponding chain reasoning prompt by each agent to analyze the operation video clip, and outputting a final error detection result based on an optimization threshold value by weighting and fusing analysis results of each agent. According to the invention, accurate detection of various surgical errors can be realized, and detection efficiency and clinical suitability are considered at the same time.

The invention discloses an offshore rocket self-lifting platform dynamic scheduling method based on cooperative intelligence, and the method comprises the following steps: constructing a plurality of platforms into an intelligent unit capable of autonomously responding to a task, and carrying out the dynamic marshalling through a cooperative mechanism, so as to achieve the self-adaptive matching between the task and the platform. In the method, multiple indexes are introduced as scheduling targets, weight parameters are dynamically adjusted according to real-time tasks, resources and environment information, and the scheduling flexibility is improved. And generating a scheduling strategy through a learning optimization mechanism, and controlling platform task allocation, path planning and execution time sequence. In the execution process, a scheduling strategy and a platform marshalling structure can be updated in time according to environment disturbance and task state changes, and smooth execution of tasks and stable cooperation of the platform are ensured. The method is suitable for a multi-task and high-frequency emission scene.

The invention discloses an unmanned aerial vehicle group dynamic path planning method and system based on cooperative intelligence, and belongs to the field of unmanned aerial vehicle intelligent control, and the method comprises the steps: obtaining a navigation task of an unmanned aerial vehicle group, and marking N task nodes for the task full period of the navigation task; for a navigation task, positioning a key navigation path point; the task path planning module is connected with the cluster path planning module, traverses N task nodes and the key flight path points, carries out cluster division on an unmanned aerial vehicle cluster according to the course, carries out task path planning by taking a node single-machine target as guidance, and determines a path planning strategy; wherein in a manner of sequential planning based on N task nodes, offset calibration is carried out based on an unmanned aerial vehicle distribution network of an upper task node. The technical problems of low cooperation efficiency, high collision risk and poor task continuity caused by delay of centralized planning response of the unmanned aerial vehicle cluster, stiffness of static grouping resources and lack of a calibration mechanism in path error accumulation in the prior art are solved.

The invention discloses a multi-agent collaborative sensing system and method, and belongs to the field of collaborative sensing. The system comprises at least two intelligent agents. Randomly selecting an agent as a main agent; the positioning module is used for transmitting the obtained positioning information to the communication module and the environment sensing module; the environment sensing module obtains a bird's-eye view feature map according to the obtained sensing data; wherein the main intelligent body transmits current positioning information to the cooperative intelligent body through the communication module, so that the cooperative intelligent body obtains a bird's-eye view feature map at a corresponding view angle according to the obtained bird's-eye view feature map; the feature compression module is used for performing spatial dimension and channel dimension compression according to the aerial view feature map and a preset spatial block to obtain a compressed feature map; the compressed feature map of the cooperative agent is transmitted to the main agent through the corresponding communication module; and the fusion sensing module is used for performing fusion according to all the received compressed feature maps to obtain a global sensing result. By implementing the method and the device, the intelligent agent sensing result can be kept at high accuracy at low bandwidth.

The invention particularly relates to an asynchronous point cloud fusion method for an air-ground cooperative sensing system, and the method comprises the steps: collecting point cloud data corresponding to a current environment through a current intelligent agent, and carrying out the coding processing of the point cloud data, so as to obtain a real-time feature tensor; receiving an asynchronous feature tensor corresponding to k frames of historical point clouds sent by the cooperative agent; wherein the cooperative intelligent agent and the current intelligent agent belong to a ground-air cooperative system; performing feature-level time synchronization processing on the current feature tensor and the asynchronous feature tensor by using a time delay compensation network to obtain a synchronous feature tensor; performing multi-scale feature weighted fusion on the synchronous feature tensors corresponding to the plurality of cooperative agents to obtain a fused feature tensor; and performing decoding processing on the fusion feature tensor to display fusion perception data corresponding to the fusion feature tensor. According to the method, the real-time performance, the accuracy and the safety of a sensing system can be improved, and the integrity and the reliability of environment sensing are improved.

The present invention discloses a multi-agent collaborative perception method under complex noise based on lidar, which includes the following steps: Step S1, perform coordinate transformation and preprocessing on the lidar point cloud of the collaborative agents; Step S2, use the method of deep learning to extract the feature maps of the lidar point clouds of all collaborative agents, and then compress the feature maps of the lidar point clouds of the collaborative agents and transmit them to the central vehicle; Step S3, the central vehicle decompresses the received feature maps of the lidar point clouds of the collaborative agents, and at the same time uses the feature maps of the lidar point clouds of a small number of previously received collaborative agents to perform corresponding deep learning processing on the transmission time delay and relative pose error; Step S4, the central vehicle uses the attention mechanism to fuse the feature maps of the lidar point clouds of the collaborative agents into the feature map of the central vehicle to obtain the final feature map; Step S5, send the final feature map into the detection head to obtain the detection result; solves the problem of the performance degradation of the multi-agent perception system under the transmission time delay and relative pose error.

The application discloses a wireless federated learning method for large-scale Internet of Things cooperative intelligence, aiming at the problem of heterogeneous device computing capability in a large-scale Internet of Things scene, and integrating centralized learning and federated learning to form a unified architecture, so that devices with weak computing capability can participate in global model training. On the one hand, the application determines the data sample selection strategy by the data importance of the centralized learning user, which can reduce the communication overhead and transmission time of data uploading; on the other hand, the application prunes the model of the federated learning user, which can effectively reduce the local computing time under the premise of ensuring the learning performance. The federated learning method provided by the application can realize data sample selection, model pruning and user scheduling of different types of users, which is helpful to improve the utilization rate of wireless network resources and alleviate the problem of limited resources of the Internet of Things.

A method, computer program product, and computing system for generating, via a video recording subsystem of an ACI calibration platform, a three-dimensional model of at least a portion of a three-dimensional space containing an ACI system; and generating, via an audio generation subsystem of the ACI calibration platform, one or more audio calibration signals for receipt by an audio recording system included within the ACI system.

The invention discloses an edge terminal-oriented memory efficient model fine tuning method and system, which realize decoupling of a backbone network and a fine tuning process by constructing a parallel side network module, and avoid video memory occupation caused by backbone gradient return. A double-adapter module is arranged, modeling is carried out on same-layer features and cross-layer context information, and the feature expression precision is improved. And constructing a feature fusion module, and performing adaptive weighting on different path outputs through learnable gating. A backbone grouping module is integrated, dynamic grouping is carried out on a backbone network based on interlayer similarity, and redundant modules and calculation overhead are reduced. And through the fine tuning execution module, only back propagation and parameter updating are carried out on the side network, and a trunk freezing state is kept, so that the memory and training cost is reduced. According to the method, the memory efficiency and the training speed of the edge fine adjustment process are improved, high-precision model self-adaption is achieved with extremely low calculation burden, and application in the fields of cloud edge collaborative intelligence and the like of unmanned aerial vehicles, robots, vehicle-mounted terminals and the like is effectively supported.

The embodiment of the invention discloses an intelligent business guiding method and device for a bank outlet, a medium and a program product, and relates to the technical field of artificial intelligence, and the method comprises the steps: determining an intention recognition agent terminal and an intention recognition task from a plurality of execution agent terminals based on the current hall environment information through an intention recognition process; sending an intention recognition task to the intention recognition agent terminal by using the collaborative planning process, so that the intention recognition agent terminal executes the intention recognition task to obtain client intention information; client intention information fed back by the intention recognition agent terminal is received by using a collaborative planning process, and collaborative agent terminals and collaborative tasks corresponding to the collaborative agent terminals are determined from the multiple execution agent terminals based on the client intention information and the agent relation graph; the corresponding cooperation task is distributed to each cooperation agent terminal by using the cooperation planning process, so that the multiple cooperation agent terminals carry out business guidance on the client, and the response speed and the client experience are improved.

The invention relates to the technical field of intelligent traffic and automotive electronics, in particular to a projection type road marking method and system based on dynamic traffic environment perception, and the method comprises the steps: collecting and fusing real-time traffic flow, meteorological conditions and event data to form an environment state data set; calculating a traffic state index to evaluate a current road traffic state; triggering a preset road marking rule based on the traffic state index, and generating a marking scheme; constructing a dynamic space mapping mechanism based on a two-dimensional coordinate system, and mapping the marking elements into coordinate data; performing geometric distortion correction and smooth transition processing on the coordinate data to obtain final marking graph data; and dynamically calculating a brightness parameter of a finally projected road marking graph in combination with a real-time environment state to finish dynamic marking projection. According to the invention, a perception-decision-projection integrated network is constructed, collaborative study and judgment are carried out based on the traffic situation of the whole road section, and evolution from isolated projection to whole-network collaborative intelligence is realized.

The invention belongs to the technical field of vehicle control, and particularly provides a data interaction method among multiple agents of a cabin, a medium and a vehicle, the agents comprise a system agent and a plurality of ecological agents, and the method comprises the following steps: obtaining capability description information predefined for each agent, the capability description information is used for declaring an executable task type and a data interaction specification of the intelligent agent; in response to generation of a target task, determining at least two cooperative agents for executing the target task based on the capability description information; establishing a secure communication link between the at least two cooperative agents, so that the at least two cooperative agents interact the task data of the target task; and based on an interaction result of the task data, collaborating the at least two collaboration agents to execute the target task. Through the technical scheme provided by the invention, the cooperation capability among multiple agents in the vehicle cabin can be improved.

The invention belongs to the technical field of water conservancy project intelligent operation and maintenance, and discloses a complex environment adaptive collaborative operation intelligent operation and maintenance robot which comprises a vehicle body, and a movable chassis system, a multi-source sensing fusion system, a multi-degree-of-freedom mechanical arm, an intelligent storage platform and an intelligent control system are arranged on the vehicle body. Through the structural design and technical innovation of all the systems, many technical problems existing in a traditional water conservancy pipe network operation and maintenance mode and existing equipment are solved in a targeted mode, and the aspects of environmental adaptability, operation functionality, mechanical stability, collaborative intellectualization, operation and maintenance economical efficiency and the like are remarkably improved; the precise sensing capability in a complex and severe environment is greatly improved, and the detection precision and the anti-interference performance are remarkably enhanced; the robot constructs a triple heterogeneous sensor fusion system of a visual camera, a first laser radar, a second laser radar and a millimeter wave radar, combines a dynamic weight distribution algorithm, gives full play to the technical advantages of each sensor, effectively inhibits the coupling influence of darkness, water mist, dirt and other multi-source interferences in a pipe network, and improves the detection accuracy. The pipe network inner wall defect detection device can accurately capture fine defects such as 0.1 cm-level cracks and corrosion on the inner wall of a pipe network, stabilizes the defect detection accuracy at 92.5% or above, fundamentally solves the problems that a single sensor or simple double sensors are low in detection precision and multiple in missed detection and false alarm in a complex pipe network environment, and is beneficial to actual application and operation.

Systems and methods for using a differentiable multi-agent Actor-Critic (DiMAC) for multi-step radiology report summarization. The tasks of extracting salient sentences and phrases are divided across two collaborating agents that are trained end-to-end using reinforcement learning (RL).

A collaborative intelligence-based semi-automated semiconductor wafer labeling system according to an embodiment of the present invention comprises: a first database providing at least one set of WBM image data that has been labeled; a second database providing a plurality of test data set control groups composed of a plurality of unlabeled WBM image data; a learning model comprising a first process for receiving WBM image data from the first database and learning defect patterns and normal patterns, a second process for predicting the defect probability of WBM image data received from the second database based on the learned pattern information, a third process for calculating an uncertainty value for each data based on the predicted defect probability value, and a fourth process for determining whether to perform an auto-labeling operation using a labeling model or an engineer labeling operation by a worker on the WBM image data according to the result of comparing the calculated uncertainty value for each data with a preset threshold value; and a first labeling unit that inputs auto-labeled WBM image data into the first database according to the determination result of the learning model and provides it as training data for the learning model. It may include a second labeling unit that inputs engineer-labeled WBM image data into a first database and provides it as training data for the learning model based on the judgment result of the learning model.

Systems and methods for task-driven collaborative smart cluster location reporting are provided. A first wireless communication entity may receive a first message from a second wireless communication entity, the first message including a collaborative smart cluster (CIC) location report configuration. The first wireless communication entity may send a second message to a second wireless communication entity, the second message including CIC location information determined based on the CIC location report configuration.

The invention discloses a man-machine cooperation intelligence agent system and method based on interactive feedback, and relates to the field related to artificial intelligence, and the method comprises the steps: receiving an intelligence analysis demand inputted by a user side, and extracting and analyzing intention elements through intention analysis; generating a task execution tree, and cooperatively adjusting the task execution tree with a user through an interactive interface; performing two-order matching routing decision based on node task characteristics in the agent cluster, determining an agent task cluster and performing parallel tree task analysis, obtaining task analysis results and integrating the task analysis results, determining a structured intelligence report and constructing an evidence tracing chain; and performing terminal interface display on the structured intelligence report and the evidence tracing chain. The technical problem that efficiency and accuracy are unbalanced in the existing intelligence analysis field is solved, and the technical effects that intelligence analysis efficiency is improved, analysis result accuracy and credibility are enhanced, and efficiency and accuracy dynamic balance is achieved are achieved.

The invention provides a traffic signal real-time optimization regulation and control system based on AI cooperative intelligence, and relates to the technical field of traffic regulation and control, and the system comprises an AI cooperative sensing module which is used for obtaining a traffic data flow; the unimpeded traffic evaluation module is used for obtaining a traffic unimpeded traffic coefficient; the coefficient judgment module is used for judging whether the traffic unimpeded coefficient is smaller than a traffic unimpeded threshold value or not; the regulation and control instruction obtaining module is used for obtaining a traffic signal regulation and control instruction; and the optimization adjustment module is used for obtaining an optimized traffic signal scheme. The technical problems of lagging traffic signal regulation and control and low traffic efficiency caused by the fact that an existing traffic signal regulation and control system cannot sense intersection traffic state changes in real time can be solved, real-time traffic data collection and analysis based on AI collaborative intelligence are achieved, the unimpeded traffic state is dynamically evaluated, the intersection traffic efficiency is improved, and the traffic quality is improved. And the traffic jam is reduced.

The invention discloses a two-dimensional magnetic material purpose drive reverse recommendation method and system, and belongs to the field of material informatics and artificial intelligence technology cross application. The method comprises the following steps: obtaining patent data and attribute data corresponding to a two-dimensional magnetic material by using a multi-source data fusion construction module, a multi-source fusion database is obtained through an ETL data flow integration processing mechanism; a multi-AI cooperation intelligent reasoning module is used for receiving a purpose demand input by a user, intelligent reasoning is carried out on the purpose demand, and a recommendation result is generated; a multi-key polling technology is realized by using a high-availability system architecture module and adopting an API polling mechanism, and a multi-level fault-tolerant processing system is constructed; the intelligent user interaction module is used for providing a response type Web interface to support a user to input a purpose demand, carrying out purpose correlation judgment and displaying an attribute combination prediction result and an intelligent recommendation result. According to the method, the problems of data splitting, unstructured semantic disorder and single recommendation mode existing in an existing material screening method are solved.

The invention discloses an intelligent agent cooperation intelligent decision-making method, system, equipment and medium for transformer substation environment control, and belongs to the technical field of intelligent operation and maintenance, and the method comprises the steps: collecting transformer substation historical data, building an intelligent agent interaction environment, carrying out the modeling of a decision-making problem, and carrying out the state estimation according to local observation and the transformer substation historical data, the method comprises the following steps: constructing a global objective function, predicting an individual contribution degree through a neural network model, defining an individual reward signal according to the individual contribution degree, calculating a return index through a primary algorithm, carrying out strategy training and optimization according to the return index, executing a decision, carrying out normalization processing on the individual reward signal, and introducing a regularization item into the global objective function. And calculating a convergence index and monitoring the convergence. According to the method, state sensing and decision making are carried out, the problems of low operation and maintenance efficiency and poor collaboration in substation control are solved, automatic decomposition and efficient decision making of a global optimization target are achieved, and the self-adaptive capacity and the overall energy efficiency of the system are improved.

The application discloses a kind of robot-assisted surgery error detection methods based on cooperative intelligent agent reasoning, comprising: constructing multi-classification surgery error system and clinical informed knowledge base;Based on multi-classification surgery error system and knowledge base, chain reasoning prompt library is constructed for different professional level and analysis perspective;To input surgery video segment, through the composite risk score of technical complexity score and clinical impact score, error instance of different risk score is distributed to corresponding professional level reasoning path;For each professional level reasoning path, time analysis agent, spatial analysis agent and process analysis agent are deployed, each agent calls corresponding chain reasoning prompt to analyze surgery video segment, and the final error detection result is output based on the analysis result of each agent by weighted fusion and optimization threshold. The present application can realize the accurate detection of multiple types of surgical errors, while taking into account the detection efficiency and clinical adaptability.

The application discloses a kind of based on collaborative intelligence's offshore rocket self-lifting platform dynamic scheduling method, comprising the following steps: multiple platforms are constructed as intelligent unit that can be autonomously responded to task, and the adaptive matching between task and platform is realized by dynamic grouping through collaborative mechanism.In the method, multiple indicators are introduced as scheduling targets, and the weight parameters are dynamically adjusted according to real-time task, resource and environmental information to improve scheduling flexibility.Through learning optimization mechanism to generate scheduling strategy, control platform task allocation, path planning and execution timing.In the execution process, scheduling strategy and platform grouping structure can be updated in time according to environmental disturbance and task state change, to ensure smooth execution of task and stable cooperation of platform.The application is suitable for multi-task, high-frequency launch scene.

The invention discloses a cooperative intelligent feature compression algorithm based on channel attention, and the algorithm mainly comprises the following steps: firstly, carrying out the feature extraction of a video frame through a feature extraction end of a YOLOX network, then selecting a splitting point, and inputting an intermediate feature at the splitting point into a feature compression network. In the feature compression network, channel aggregation based on channel attention is firstly used to carry out redundancy elimination on feature channel dimensions, then convolution is carried out to remove spatial redundancy, and finally, compression features are recovered at the cloud and input into the rear half part of the YOLOX network to obtain a video target detection result. A generalized division normalization layer is introduced in the feature compression and reconstruction process to replace a conventional batch normalization layer, so that the features are reconstructed with higher precision. Experimental results show that useful information can be completely reserved during channel fusion, communication bandwidth pressure is effectively relieved, and machine vision task performance is guaranteed.

The invention provides an intelligent manufacturing process management and control method based on an AI Agent. The method comprises the steps that firstly, a task set is constructed, and the AI Agent with sensing, decision making and state maintenance capabilities is initialized; secondly, carrying out attribute modeling on various manufacturing resources, and establishing a task-resource capability function; then, a process state transition diagram and a strategy model are constructed, and task scheduling path reasoning is achieved; then, the process income is evaluated through a unified objective function, and the contribution of each Agent is evaluated; and finally, when the local income is lower than a threshold value, triggering flow reconstruction, and supporting reconstruction adjustment of the flow under resource change or emergencies, thereby realizing intelligent perception, autonomous decision and dynamic optimization control of the manufacturing flow. According to the invention, an extensible, explainable and optimizable intelligent process management infrastructure is provided for the intelligent manufacturing system, and the operation efficiency, response flexibility and collaborative intelligence level of the manufacturing system can be remarkably improved.

The invention belongs to the technical field of energy management, and particularly relates to a source network load storage energy management system which comprises a new energy power generation prediction module, a distributed distribution network intelligent protection module and an optimal scheduling module. The power generation prediction module is used for realizing high-precision power generation output prediction by collecting environment type and equipment type power generation influence parameters and utilizing a deep learning algorithm fused with a space-time attention mechanism; the intelligent protection module calculates an output fluctuation coefficient and a voltage deviation ratio by collecting voltage, current and power flow direction data of key nodes of a distribution network, dynamically identifies bidirectional power flow impact and voltage out-of-limit risks and executes hierarchical pre-control; and the optimal scheduling module performs fitting and supply-demand relationship analysis on the source network load storage time sequence data based on an LSTM and a multi-model fusion algorithm, and realizes adaptive dynamic scheduling of energy flow. According to the method, new energy output prediction accuracy, distribution network risk prevention and control initialization and source network load storage collaborative intelligence are realized, and the safety and economical efficiency of system operation are remarkably improved.