223 results about "Branch node" patented technology

The invention discloses a power transmission line fault positioning method and system based on ground potential end transient information perception, and relates to the technical field of line fault positioning, and the method comprises the steps: obtaining a voltage traveling wave signal and a current abrupt change rate signal detected by a trunk line and branch node sensor array in a power transmission line; processing the voltage traveling wave signal and the current abrupt change rate signal to obtain traveling wave components and polarity-amplitude calibration results of different modes; according to traveling wave components and polarity-amplitude calibration results of different modes, in combination with topological parameters of the power transmission line and spatial distribution of a ground potential end sensor, adjusting edge weights or connection relationships in the graph structure to obtain a composite feature matrix and an edge weight matrix of each node; and inputting the composite feature matrix and the edge weight matrix of each node into a graph neural network positioning model, and outputting a power transmission line fault positioning result. According to the invention, through ground potential end information fusion and the dynamic graph neural network, accurate fault positioning under a complex power transmission topology is realized.

A system and a method for optimizing a cable system in a telecommunication network Includes the steps of: receiving a position of each of a plurality of terminal nodes and a minimum bandwidth requirement associated with an operation of the telecommunication network including the plurality of terminal nodes and a plurality of branching nodes, wherein the telecommunication network is arranged to facilitate communication among the plurality of terminal nodes via at least one of the plurality of branching nodes; determining, between adjacent pairs of nodes of terminal nodes or branching nodes, a distance and a bandwidth capacity of a cable edge between each adjacent pair; and providing an optimized geographical location of the branching nodes and an optimized bandwidth capacity of each cable edge in which the telecommunication network being deployed with the branching nodes and the cable edges meets a quality of service associated with the minimum bandwidth requirement.

The invention relates to the technical field of database management, and discloses a hybrid memory index management method and device, equipment and a medium. Receiving an access request, providing a target index node from the cache when the access request is hit, and reading the target index node from the nonvolatile storage medium when the access request is not hit; writing branch nodes into a high-bandwidth memory cache according to node types, writing leaf nodes into a dynamic random access memory cache, and registering information; when the cache of the high-bandwidth memory meets a first trigger condition, migrating the branch node to the cache of the dynamic random access memory and updating information; and when the cache of the dynamic random access memory meets a second trigger condition, migrating the branch node to the high-bandwidth memory cache, replacing the node and updating the information. According to the method, the node residence position is dynamically adjusted by distinguishing the node types and combining the cache performance, and the index node mapping table is synchronously updated, so that the index request processing performance and the storage resource utilization efficiency are improved.

The invention relates to the technical field of path planning of carrying crown blocks, in particular to a path re-planning method based on accurate cost calculation of an AMHS system. The method comprises the following specific steps: S1, acquiring a real-time traffic condition in a current state in an AMHS system; s2, calculating the path cost of each OHT crown block at the current moment based on the obtained real-time data; and S3, when the OHT crown block runs to the branching point, according to the path cost calculated in the step S2, selecting the path with the minimum current cost in the plurality of alternative paths in a dynamic change manner. According to the method, on the basis of multiple existing alternative paths, data such as the number of vehicles, the speed of the crown blocks, the distance between the crown blocks and the working state in the alternative paths are obtained based on the real-time traffic condition of the AMHS system, and the path cost at the current moment is accurately calculated through a cost formula designed in the technical scheme; and it is ensured that the crown block can select the path with the minimum current cost from the multiple alternative paths with the dynamically-changed cost at the branch node.

The invention discloses a target model interaction method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining a reasoning instruction; performing data reasoning based on the reasoning instruction through the target model to obtain reasoning process data and a reasoning result; wherein the reasoning process data comprises a plurality of branch nodes and a plurality of conclusion nodes which are cascaded in a data reasoning process; and displaying the reasoning process data and the reasoning result. Through the mode, the transparency and the interpretability of the reasoning path of the target model can be improved, and the efficiency of generating the model output result meeting the user expectation by the target model is improved.

The invention relates to the technical field of network communication and video processing, in particular to an intelligent global cloud live broadcast system based on AI and multi-operator and multi-cloud optimization, and the system comprises a data collection module which is used for collecting live broadcast video streams; the video compression module adopts H.265 coding; the content review module is used for performing multi-modal review; the RPMTS packaging module is used for accessing three operator cloud private line POP nodes through a public network after packaging; the decoding module is used for decoding POP nodes in the middle area; the translation module is used for generating multi-language subtitles or dubbing; the multi-operator and multi-cloud adaptability matrix module calculates an optimal path in combination with an AI algorithm and a multi-cloud cost adaptability matrix; the multi-branch access module is used for realizing branch node cooperative forwarding; and the multi-language distribution module is used for generating multi-language subtitles and dubbles and pushing the multi-language subtitles and dubbles to each global live broadcast platform, so that synchronous distribution is realized, and the system can reduce global live broadcast delay, improve multi-language coverage capability and realize high reliability and cost optimization.

The invention provides a digital twin unmanned aerial vehicle circuit state monitoring method and system, and the method comprises the steps: firstly collecting a partial discharge pulse sequence of an unmanned aerial vehicle switching power supply circuit, separating environment electromagnetic noise through polarization filtering operation, and extracting real discharge pulses to generate partial discharge fingerprint data; inputting the fingerprint data into a pre-established fault search tree model, judging the type of a capacitor or inductor component through a first-stage branch node, and matching a degradation mode such as capacitance attenuation or magnetic saturation through a second-stage branch node to determine the fault type; then, corresponding component parameters of the equivalent circuit model are dynamically updated in the digital twin body, and the circuit operation state change trend is simulated; and finally, according to the state change trend, predicting the residual effective working time of the fault component and outputting a monitoring result. According to the invention, the accuracy of service life evaluation of key components in a complex electromagnetic environment is improved.

The invention discloses a gene identification method based on a mask graph auto-encoder. The method comprises the following steps: 1, acquiring gene interaction data and omics characteristic data and preprocessing the data; 2, masking the processed data, specifically comprising two branches, a node masking module and an edge masking module; 3, inputting the masked network into a graph automatic encoder for training, wherein the graph automatic encoder learns embedded representation of the network by reconstructing nodes and edges of the network; and 4, obtaining low-dimensional embedding of the features through the trained encoder, and finally performing gene classification by using a logic regression classifier. According to the method, the node information and the structure information of the graph are concerned at the same time by respectively masking the nodes and the edges in the network, and the dependence of the feature training model on the label information is reduced in a self-supervised learning mode, so that the genes can be accurately classified.

The invention discloses an industrial chain generation method and system based on big data, and the method comprises the steps: designing a main structure of an industrial chain and a keyword table of each node, and determining the classification of the upstream, the middle and the downstream of the industrial chain, including a main node, a child node, a branch node, a name of each node and a keyword contained in the node; multi-dimensional industrial data such as industry and commerce data, bidding and tendering, public opinions, patents and software aboriginal information of enterprises are collected and acquired, and an enterprise basic database is established; primary screening of the uplink enterprises: screening the enterprises from the enterprise basic database according to the industrial chain node keywords, and incorporating the enterprises into an alternative uplink enterprise database; the enterprise upper chain evaluation index system is used for evaluating the matching degree between dimension information such as the business scope, the technical achievement, the public opinion information and the bidding and tendering of the enterprise and the industrial chain node; selecting each node benchmark enterprise on an industrial chain according to industrial research, and calculating each uplink index of each benchmark enterprise; calculating a vector distance between the primarily screened selected enterprise and the benchmarking enterprise; ranking the similarity calculation results in a descending order; according to the Pareto's law, the first 20% of enterprises are taken to be linked, and a complete industrial chain is constructed. The method comprises the following steps: collecting enterprise multi-dimensional data, constructing an industrial chain main body structure and each node keyword, screening to-be-linked enterprises according to keywords, establishing an uplink evaluation index system, calculating vector distances between the to-be-linked enterprises and benchmarking enterprises, and determining an uplink enterprise mode according to vector distance ranking to generate an industrial chain. The method is beneficial for improving the accuracy and efficiency of industrial chain generation, so that reference basis is provided for scientifically evaluating the industrial development level, making investment decisions of industrial development, making effective industrial policies and the like.

The invention discloses a transformer area topology identification method and device based on multi-feature fusion, and the method comprises the steps: S1, distributing an initial branch node for each user, constructing an initial topological structure identification model, and obtaining a preliminary branch node-user association relation adjacency matrix; s2, constructing a topological optimization model based on difference characteristics, and enabling the error between branch node power difference and user power difference combinations to be minimum; s3, constructing a topological optimization model based on peak-valley feature matching, and enabling the matching degree of the peak-valley feature combination of the branch node and the peak-valley feature combination of the user to be maximum; s4, constructing an optimization model based on total fitting error correction, and enabling the total error between the actual power and the fitting power of the branch node to be minimum; s5, judging whether undistributed users exist or not, if not, obtaining a final topology identification result, and otherwise, performing S1 to S4 again; according to the method, the area topology connection relation is effectively identified.

The invention provides a power distribution network single line diagram automatic generation method and system, and belongs to the technical field of power distribution network single line diagram automatic layout, and the method comprises the steps: determining a plurality of connection relation lines based on power grid topology data, and determining a trunk line based on the plurality of connection relation lines. The trunk line includes a plurality of main line nodes. And determining a primary branch line path corresponding to each main line node based on the main line. The primary branch line path includes a plurality of branch nodes. And determining a second-level branch line path corresponding to each branch line node based on the first-level branch line path. And generating a power distribution network single line diagram based on the trunk line, the first-stage branch line path and the second-stage branch line path. According to the automatic generation method and system of the power distribution network single line diagram, the drawing efficiency of the power distribution network single line diagram can be improved.

The invention relates to a broadband power grid oscillation positioning method and device based on modal analysis. The method comprises the steps of obtaining broadband oscillation data in a power grid, extracting frequency spectrum characteristics of the broadband oscillation data, identifying an oscillation mode based on the frequency spectrum characteristics, determining an oscillation positioning index of each branch node in the power grid based on the broadband oscillation data, and determining the frequency spectrum characteristics of each branch node in the power grid based on the oscillation positioning index of each branch node. And constructing a spatial-temporal feature tensor containing a time feature and a spatial feature, adding the identified oscillation mode as a tag to the spatial-temporal feature tensor, and inputting the spatial-temporal feature tensor to a trained oscillation source positioning model to obtain an oscillation source positioning result. The oscillation source positioning model is obtained by training a long short-term memory network based on a historical spatial-temporal feature tensor carrying an oscillation source position label. By adopting the method, the oscillation source can be efficiently and accurately positioned.

The invention relates to a method and a system for accelerating large model reasoning by using dynamic tree structure speculative decoding. The method comprises the following steps: processing input characters by adopting a small model to generate prediction characters of a tree structure; inputting the input characters and the prediction characters generated by the small model into the large model, obtaining an output result of the large model through a large model reasoning process, and judging whether the prediction characters generated by the small model are received or not, so as to obtain a finally generated text. According to the tree reasoning structure, the branch nodes with higher expected receiving probability can be utilized, and more tokens are expected to be generated while the speculation length is increased, so that the purpose of accelerating large model reasoning is achieved; the design of the dynamic determination ending condition can be ended more quickly when the method is poor in performance, and the method has better comprehensive performance compared with similar methods.

The invention discloses a container isolation and double-node management fused data integration governance method and system, and relates to the technical field of data integration governance. Communication time sequence data between a plurality of groups of main nodes and branch nodes of a hydropower station is monitored in real time, and deep analysis is performed by using a neural network model CNN; indexes such as data drift rate and data coverage difference can be accurately identified, and the data synchronization condition can be evaluated in real time. The accuracy of data synchronization between the nodes is promoted to be improved, and the data consistency and timeliness are ensured. By constructing the communication state model and deeply analyzing the data, the communication state can be accurately evaluated, and key communication indexes such as the uplink and downlink transmission ratio, the burst load index and the like can be identified. Once it is found that the communication state is unstable, the first governance strategy, the second governance strategy and the third governance strategy can be triggered, so that the data transmission path is optimized, the network load is adjusted, resource allocation is effectively optimized, and the overall operation efficiency of the system is improved.

The invention is suitable for the technical field of aircraft obstacle avoidance, and particularly relates to an aircraft obstacle avoidance control method and system, and the method comprises the steps: delimiting a flight region of an aircraft, collecting the stereoscopic optical remote sensing data of the flight region, drawing an obstacle recognition graph, marking the position of an obstacle, and configuring the color of the obstacle. Wherein one obstacle at least corresponds to one color, calculating the height of each obstacle, and marking the height in the obstacle identification graph; and segmenting the height into a plurality of layers, and creating branch nodes and leaf nodes. By constructing the obstacle avoidance tree, the color data of all obstacles can be visually displayed, and the color image can be quickly verified, so that the flight safety is greatly improved while the normal flight of the aircraft is ensured, the suspicious obstacles on the flight path can be verified by extracting the RGB values, the flight path planning is optimized, and the flight path planning efficiency is improved. And the flight safety is further improved.

The invention discloses a gas pipe network simulation analysis system. Comprising a gas pipe network data acquisition unit, a pipe network topological structure building unit, a pipe network data value influence factor analysis unit, a gas pipe network data prediction model building unit, a gas pipe network data periodic prediction unit, a gas pipe network data regulation and control unit, a risk early warning unit and a gas pipe network data storage unit. According to the method, the simulation of the gas pipe network is realized by building the pipe network topological structure in the virtual environment; meanwhile, periodic prediction of the gas pipe network data is achieved through a time recurrent neural network model, the collected real data value on the gas pipe network is compared with the periodic prediction range value of the gas pipe network data, the risk of the gas pipe network is judged, and if the real data value exceeds the periodic prediction range value of the gas pipe network data, the risk of the gas pipe network is judged. And if yes, regulating and controlling the real data value of the gas pipe network data exceeding the preset threshold value by regulating valves on branches and branch nodes in each constituent part in the virtual pipe network.

The invention discloses a question and answer intention recognition and classification method for an infant care robot. A question rewriting node, an intention recognition node and a branch node are included. The input data of the question rewriting node are questions input by a user, the training data of the question rewriting node are phrases for mothers, infants and child rearing scenes, the model parameters of the question rewriting node are obtained through adjustment based on the training data, and the question rewriting node processes the input data according to the model parameters of the question rewriting node and outputs the data; the intention recognition node obtains data output by the problem rewriting node and recognizes a user intention and an expected result; the branch nodes provide a plurality of large language models in mother and child breeding scenes, and the branch nodes select the corresponding large language models to execute data processing according to output data of the intention recognition nodes. Original user input is clarified through standardization and rewriting, then the accuracy and response efficiency of subsequent intention recognition nodes are remarkably improved, and finally the overall user experience is improved.

The invention relates to a traveling wave device stationing and fault positioning method suitable for a load transfer operation mode, and belongs to the technical field of power system fault distance measurement. The method comprises the following steps: obtaining a branch forest in a normal operation mode based on a line topological structure; obtaining a branch forest in a load transfer operation mode according to the load transfer operation of the two connection lines; if the topological tree structures and leaf nodes of the two branch forests under the same root node change, preferentially arranging traveling wave devices at the leaf nodes appearing in the two different topological trees under the same root node, otherwise, calculating the weight coefficient of each branch in the topological tree structures, and arranging traveling wave devices at the branch nodes with high weight coefficients; when a single-phase earth fault occurs on the line, the distance between each traveling wave device and the fault point is calculated, fault section tracing is carried out, and the position of the earth fault point is determined. The objective of the invention is to solve the technical problem that the prior art cannot adapt to a new topological structure after load transfer.

The invention discloses a question answering method and device, electronic equipment and a computer program product. The method comprises the steps that a to-be-replied question is acquired, the to-be-replied question at least comprises management information and to-be-replied question data, and the management information at least comprises multiple levels of management grids for managing the to-be-replied question; in a plurality of knowledge nodes of a preset question and answer knowledge tree, a management path matched with the management information is determined, historical question data and corresponding historical reply data are pre-configured in each preset leaf node, and the management path comprises a plurality of branch nodes arranged according to the level sequence of management grids; in the at least one first leaf node, screening a target leaf node matched with the question data to be replied; and determining the historical reply data in the target leaf node as question reply data of the to-be-replied question. According to the invention, the technical problem of high computing resource consumption in a question answering mode in the prior art is solved.

The invention discloses a multi-role branch chain node state linkage management method and system, and belongs to the field of node collaborative management, and the management method comprises the following specific steps: S101, setting role nodes, adding attribute information of each role node, setting a node basic identification rule, and generating a branch chain according to an intervention approach; the method ensures that different role node numbering systems are independent and not confused, fundamentally avoids the problems of serial number repetition, vacancy or logic fracture, avoids branch structure chaos after role node upgrading, ensures that the hierarchical relationship of nodes in branches before and after conversion is coherent, avoids misadjustment of irrelevant role nodes during exception handling, and reduces system resource consumption. The state adjustment efficiency is improved, manual operation and maintenance links are reduced, manual operation errors are avoided, and meanwhile the response speed of node management and state adjustment is improved.

The invention relates to the technical field of government affair system management, in particular to a cross-unit cross-level approval management method. Comprising the following steps: S1, obtaining an approval item and an approval attachment submitted by an initiator, and determining one or more final approval objects of the approval item according to a pre-configured approval authority; s2, obtaining a level and a unit where an initiator is located and a level and a unit where a final approval object is located, and determining a transfer node of the approval in combination with a pre-configured organizational structure; s3, identifying the approval attachment submitted by the initiator through an OCR identification technology, and judging whether the approval attachment conforms to a specification or not; and S4, when the approval attachment conforms to the specification, determining a flow path according to each flow node of the approval, when a plurality of final approvers exist, determining the flow node as a branch node to generate a plurality of flow paths, sending the approval to each flow node, and obtaining the approval opinion of each flow node.

Provided herein is a method and system for generating and instantiating an augmented tree model from an artificial intelligence (AI) model via a Parallel and Split Learning (PSL) controller. This method receives an AI model including input and output layers, acquires computing resources of various network nodes, and obtains link information of the numerous links connecting these nodes. It segments the AI model into different partitions, thus creating an augmented tree model featuring a root node for the top-level partition, leaf nodes for the bottom-level partition, and branch nodes for the middle-level partition. These nodes are organized in a tree structure with each branch node linked to several leaf nodes or the root node linked to multiple branch nodes. The method also includes the instantiation of one node from the augmented tree model into an entity of the network, based on matching computing resources, link information, and communication requirements.

The invention provides an anti-registration error graph interaction remote sensing image change detection method and equipment. The method comprises the following steps: respectively extracting local texture features and a spatial dependency relationship of a dual-temporal remote sensing image through a CNN branch and a GCN branch which are arranged in parallel; the CNN features and the GCN features of the same time phase are fused, the two kinds of features are projected into graph structure representation, cross-branch node feature interaction is achieved through an attention mechanism, and fusion features are generated through gating weighting; inputting the dual-time-phase fusion features into an adaptive neighborhood graph sensing module, dynamically constructing an interactive edge relationship between cross-time-phase nodes, jointly updating edge features and node features to model a neighborhood change mode, and outputting a change feature graph; and generating a final change detection result according to the change feature map. By introducing a structure modeling capability and a dynamic message passing mechanism in the graph neural network, image registration error interference is effectively relieved, and precision, robustness and complex ground feature layout adaptability of remote sensing image change detection are remarkably improved.

The invention relates to the field of communication power supply and distribution management, and discloses a communication power supply and distribution monitoring system, which comprises an edge sensing system, a carrier communication network and a master station monitoring platform, greatly improves the performance and reliability, and guarantees the stability of communication services. The edge sensing system is deployed at a branch power distribution cabinet, a tail end cabinet and an equipment load, can accurately collect voltage, current, frequency and other electric power parameters in real time, and refines the monitoring granularity to a single device. The carrier communication network uses a power line as a data transmission medium and uses a carrier communication technology, so that the construction cost is reduced by using existing facilities, and load side data can be stably and timely transmitted to a master station through dynamic signal adjustment. The master station monitoring platform integrates multiple functions and can centrally manage the operation state of the system. The three devices work cooperatively to achieve full-link panoramic monitoring from the branch nodes to the single device, thereby making up for the defects of the prior art and improving the efficiency of the communication power supply and distribution monitoring system in all directions.

The invention discloses an intention recognition processing method and device based on a learning machine and a terminal, and belongs to the technical field of artificial intelligence. The method comprises the steps of obtaining request information input by a user, converting the request information into text information, and performing data cleaning to obtain cleaned text information; intention recognition analysis is carried out, and standard scene processing matching is carried out through a preset rule; meanwhile, in combination with the deep semantic understanding ability of a large language model, intelligent analysis of user intentions is carried out on the cleaned text information, a multi-dimensional optimization process is started for different analyzed branch paths, and optimized branch nodes are obtained by dynamically adjusting algorithm parameters and resource allocation; and performing context relevance check and historical behavior pattern analysis on the optimized branch nodes, generating an execution path considering both efficiency and accuracy, and obtaining an intention recognition result through the execution path. According to the intention recognition method, the expression of the user is understood by combining a large model, and the recognition accuracy is greatly improved.

The invention discloses a vehicle data query method and a related device, a cloud server is divided into a branch node and a coordination node, the coordination node does not store vehicle data and is responsible for query request verification, authority token generation and query result forwarding, and the branch node stores data and is responsible for query request verification. In the whole query process, verification permissions are not concentrated on a single node, multiple verification of identities and permissions is achieved, the whole vehicle query process is encrypted, the query result is further verified at a user side, and a set of high-security data query method is achieved.

The invention relates to the field of power distribution network black start, in particular to a hybrid V2G facility driven power distribution network grading black start control method, which comprises the following steps: after a configuration network loses power, judging whether a V2G facility is in place or not, if so, setting the selection priority of a grading black start power supply, and if not, setting the selection priority of the grading black start power supply; wherein the black-start power supply comprises a V2G facility, a diesel generator, a distributed new energy source and a thermal power generating unit; sequentially entering first-stage black start, second-stage black start and third-stage black start, and implementing black start control strategies of corresponding stages; if all the target nodes recovered at the current level are recovered, entering next-level black start; wherein the control targets of the first-stage black start, the second-stage black start and the third-stage black start are as follows: focusing a core looped network branch node, and ensuring that a core load and a power supply auxiliary machine supply power; main branch nodes are expanded, and part of residential areas and industrial and commercial areas are recovered; and carrying out multi-source collaborative frequency modulation and global grid-connected recovery. Compared with the prior art, the method has the advantages of realization of graded black start, rapid and stable recovery and the like.

The invention discloses a multi-dimensional low-voltage transformer area theoretical line loss prediction method, which comprises the following steps of: acquiring and calculating a three-phase loop impedance and three-phase current N-point curve data matrix by using electricity acquisition equipment, and calculating live line impedance values and zero line impedance values of an electric meter, a meter box and a branch node; calculating to obtain a zero line daily current square N point curve data matrix; and calculating the theoretical line loss of the current low-voltage transformer area, wherein the theoretical line loss comprises all electric meter fixed theoretical loss, all meter front theoretical line loss, all meter box-to-branch theoretical line loss and all branch-to-transformer theoretical line loss. According to the multi-dimensional low-voltage transformer area theoretical line loss prediction method, multi-dimensional data is adopted, so that the theoretical line loss prediction precision is effectively improved; different data qualities can be effectively adapted, and the adaptability is high; the theoretical impedance of each section of line in the low-voltage transformer area can be calculated in a segmented manner, and more refined line loss problem analysis is realized.

The invention relates to the technical field of data processing, in particular to a distributed intelligent customer service database updating method, which comprises the following steps of: based on configured customer service dialogue data, matching an initial node of a fault tree; according to the customer service dialogue data corresponding to the initial node, dividing process nodes, and generating an execution state label by utilizing the rule type and the text content of each process node; based on the execution state label of the current process node, combining the rule type violation record and the branch parameter of the current process node, and generating a communication portrait corresponding to the process node; comparing the intention identification deviation between each process node and the adjacent process node by using the communication portrait of each process node, and correcting the abnormal branch node according to the branch node corresponding to the intention identification deviation; performing node association analysis on the corrected branch nodes, and adjusting the branch positions of the branch nodes in the fault tree; the efficiency and accuracy of updating the customer service database are realized.

The invention relates to the technical field of path planning, and discloses a path optimization method for multi-vehicle circulating goods taking three-dimensional boxing, a terminal and a medium. The method comprises the following steps: acquiring related data, uniformly mapping part requirements of suppliers into a box body set taking a packaging box as a unit, and determining a unit residual space opportunity cost coefficient; constructing an original model according to the related data, wherein a target function of the original model comprises vehicle driving cost, vehicle starting fixed cost and compartment remaining space opportunity cost; a Dantzigzag-Wolfe decomposition method is adopted to decompose the original model into a limiting main problem and a pricing sub-problem; iteratively solving a limitation main problem and a pricing sub-problem through a column generation algorithm; and if the solution for limiting the main problem contains a non-integer, entering a branch and bound process, repeatedly executing a column generation algorithm on branch nodes until a preset termination condition is met, and outputting an optimal circulating goods taking scheme including vehicle path planning and carriage three-dimensional boxing layout. According to the invention, the transportation efficiency of circulating goods taking can be improved.