Search

CN-121980530-A - Sports fitness equipment management method based on artificial intelligence algorithm

CN121980530ACN 121980530 ACN121980530 ACN 121980530ACN-121980530-A

Abstract

The invention relates to the technical field of intelligent sports equipment management, and discloses a sports fitness equipment management method based on an artificial intelligence algorithm. The method comprises the steps of merging static description texts and dynamic use logs of equipment to generate a mixed information stream, carrying out mode slicing on the mixed information stream, separating structural feature slices and functional feature slices of the equipment, respectively excavating a static topological relation network, capturing a dynamic behavior track sequence, carrying out space-time alignment to form a joint characterization vector, constructing a relation diagram model taking the equipment as a center, taking nodes as equipment entities, enabling edges to contain static and dynamic relations, traversing the edges in the model, identifying and aggregating edge clusters with the same relation mode, abstracting each edge cluster into a management dimension, and automatically constructing a multidimensional management framework. The method realizes the automatic discovery of the management dimension from the multi-source data, and improves the intelligent level of management.

Inventors

  • GAO FANGFANG
  • GUO LEI

Assignees

  • 延安大学西安创新学院

Dates

Publication Date
20260505
Application Date
20260122

Claims (10)

  1. 1. The physical fitness equipment management method based on the artificial intelligence algorithm is characterized by comprising the following steps: fusing static description text and dynamic use log of the sports fitness equipment to generate a mixed information stream; Performing mode slicing on the mixed information flow, and separating structural characteristic slices and functional characteristic slices of equipment; excavating a static topological relation network of equipment from the structural feature slice, and capturing a dynamic behavior track sequence of the equipment from the functional feature slice; performing space-time alignment operation on the static topological relation network and the dynamic behavior track sequence to form a joint characterization vector of equipment; Constructing a relationship graph model taking the equipment as a center by utilizing the joint characterization vector of the equipment, wherein nodes of the relationship graph model comprise equipment entities, and edges comprise static topological relations and dynamic behavior associations; and traversing all edges in the relational graph model, identifying and aggregating edge clusters with the same association mode, and abstracting each edge cluster into a management dimension to obtain a multidimensional management framework.
  2. 2. The method for managing physical fitness equipment based on an artificial intelligence algorithm according to claim 1, wherein the static description text includes specification parameters, material composition and installation configuration information of the equipment, the dynamic usage log includes a running time record, a usage frequency record, a maintenance operation record and a fault alarm record of the equipment, the mixed information stream is a continuous data record formed by interweaving and splicing the static description text and the dynamic usage log at a time stamp or a logical association point, the structural feature slice is a segment set extracted from the mixed information stream and used for describing physical composition and connection relation of the equipment, the functional feature slice is a segment set extracted from the mixed information stream and used for describing running state and interaction behavior of the equipment, the static topological relation network is a network relation graph formed by connection and space layout dependence of physical components of the equipment, the dynamic behavior track sequence is a behavior record chain arranged in time sequence and used for describing state change and event triggering of the equipment, the joint characterization vector is a numerical representation fused with static network feature and dynamic sequence feature, the relation graph is a multi-dimensional cluster management model, and the multi-dimensional equipment is a multi-dimensional system is defined by using a multi-dimensional relation management model.
  3. 3. The method for managing sports equipment based on an artificial intelligence algorithm according to claim 1, wherein the merging the static descriptive text and the dynamic usage log of the sports equipment to generate the mixed information stream comprises: respectively acquiring the static description text and the dynamic use log from an equipment registration database and an equipment operation monitoring system; marking logic time labels for each item in the dynamic use log, and marking logic configuration labels for each item of parameter in the static description text; Pairing the logic time tag records with the association relationship with the logic configuration tag parameters to form a pairing record group; Sequentially splicing the dynamic use record text and the static description parameter text in each pairing record group, and inserting a separation marker at the spliced part to form a mixed information tuple; And connecting all the mixed information tuples in series according to the sequence of the logic time labels to form the mixed information stream.
  4. 4. The method for managing sports fitness equipment based on artificial intelligence algorithm according to claim 1, wherein the performing mode slicing on the mixed information stream to separate structural feature slices and functional feature slices of the equipment comprises: Setting a structure recognition mode library and a function recognition mode library, wherein the structure recognition mode library comprises text modes for describing equipment parts, connection and materials, and the function recognition mode library comprises text modes for describing equipment starting, running, stopping and alarming; Sequentially scanning the mixed information flow, extracting the text fragment and the context thereof when the scanned text fragment is successfully matched with any one of the structure recognition mode libraries, marking the text fragment and the context thereof as a structural feature slice, and recording the start and stop positions of the slice in the information flow; When the scanned text segment is successfully matched with any one of the function recognition mode libraries, extracting the text segment and the context thereof, marking the text segment as a function characteristic slice, and recording the start and stop positions of the slice in an information stream; And respectively stripping all marked structural feature slices and functional feature slices from the mixed information stream according to the start-stop positions to form independent structural feature slice sets and functional feature slice sets.
  5. 5. The method of claim 1, wherein mining the static topological relation network of equipment from the structural feature slice comprises: Analyzing each structural feature slice, and identifying the names of the equipment components mentioned in the slices and the connection relation descriptors among the components; constructing a preliminary local topology subgraph by taking each identified equipment component name as a network node and connecting relation descriptors among components as directed edges among the nodes; merging the nodes of the same-name parts described in the different structural feature slices, and aggregating all edges connected to the same-name parts; And analyzing the overlapped nodes among the partial topological sub-graphs, and splicing and fusing a plurality of partial topological sub-graphs based on the overlapped nodes to form a global static topological relation network which comprises all components and connection relations.
  6. 6. The method of claim 1, wherein capturing a sequence of dynamic behavior trajectories of equipment from the functional feature slice comprises: sorting the functional feature slice sets according to time sequence, and sorting according to event time stamps or log sequence numbers recorded in each functional feature slice; extracting core action verbs and state description words from each sequenced functional feature slice to form action state units; Arranging the extracted behavior state units according to the sequence of original functional characteristic slices to form an original behavior chain describing continuous behavior change of a single equipment; And identifying and marking key nodes representing state transition in the original behavior chain, compressing behavior state units between adjacent key nodes, and generating the simplified dynamic behavior track sequence taking the key states as nodes.
  7. 7. The method for managing sports fitness equipment based on artificial intelligence algorithm according to claim 1, wherein performing a space-time alignment operation on the static topological relation network and the dynamic behavior track sequence to form a joint characterization vector of equipment comprises: Generating a structure embedded vector for each node in the static topological relation network, wherein the structure embedded vector encodes the connectivity, centrality and adjacent node attributes of the node in the network; generating a behavior embedding vector for each state node in the dynamic behavior track sequence, wherein the behavior embedding vector encodes the occurrence frequency and the front-back state transition probability of the state node; Establishing a semantic mapping relation table between the component nodes in the static topological relation network and the state nodes in the dynamic behavior track sequence; For each node pair with a corresponding relation in the mapping relation table, splicing the structure embedded vector and the behavior embedded vector to obtain a fusion vector of the node pair; And carrying out weighted average processing on the fusion vectors of all the node pairs, and finally outputting the joint characterization vector representing the integral characteristics of the equipment.
  8. 8. The method for managing physical fitness equipment based on the artificial intelligence algorithm according to claim 1, wherein constructing an equipment-centric relationship graph model using the joint characterization vector of the equipment comprises: taking each piece of sports fitness equipment as a graph node, wherein the initial characteristic of the graph node is the joint characterization vector of the equipment; Calculating a similarity value between joint characterization vectors of any two devices, wherein two device nodes with the similarity value exceeding a preset threshold value are connected by one edge, and the weight value of the edge is the similarity value; Analyzing the shared component or physical connection relation of the cross-equipment existing in the static topological relation network, if the shared component or direct physical connection exists between two pieces of equipment, adding an edge between two corresponding nodes in the relation diagram model, and endowing the edge with a type label representing structural association; Analyzing cooperative use modes or time association of the cross-equipment existing in the dynamic behavior track sequence, adding an edge between two corresponding nodes in a relation graph model if the two pieces of equipment are used cooperatively or continuously in time, and giving a type label representing behavior association to the edge; And integrating all nodes, similarity edges with weights, structure association edges with type labels and behavior association edges to complete the construction of the relational graph model.
  9. 9. The method of claim 1, wherein said traversing all edges in the relational graph model, identifying and aggregating clusters of edges having the same association pattern comprises: extracting attributes of all edges from the relation graph model, wherein the attributes comprise weight values of the edges, type labels of the edges and node feature vectors for connecting two ends; defining a correlation mode, wherein the correlation mode is defined by a specific edge type label combination, a weight value range and a similarity range of node characteristics at two ends; comparing each edge in the relation graph model with all the defined association modes, and marking the edge as an association mode instance if the attribute of one edge completely accords with the definition condition of a certain association mode; And classifying all the edge instances marked as the same association mode into the same set to form edge clusters, and distributing unique management dimension identification for each edge cluster.
  10. 10. The method for managing physical fitness equipment based on the artificial intelligence algorithm according to claim 1, wherein the abstracting each side cluster into a management dimension, obtaining a multidimensional management framework comprises: Analyzing, for each edge cluster, a set of fixture nodes to which all edges within the edge cluster are connected, classifying all nodes in the set of fixture nodes into an independent management dimension defined by the edge cluster; recording an association mode description corresponding to each management dimension and an equipment node identification list belonging to the management dimension; The management dimensions defined by all the edge clusters are summarized to form the multi-dimensional management framework comprising a plurality of independent and orthogonal management dimensions, wherein each dimension provides a view for classifying, retrieving and managing sports fitness equipment.

Description

Sports fitness equipment management method based on artificial intelligence algorithm Technical Field The invention relates to the technical field of intelligent sports equipment management, in particular to a sports fitness equipment management method based on an artificial intelligence algorithm. Background The intelligent management of the current sports fitness equipment mainly adopts a technical path for independent analysis of a single data source. Static attribute information is generally stored in an asset management database in a structured form to form a relatively closed asset information system, and dynamic use data is mainly used for generating basic use frequency statistics or realizing a simple threshold alarming function in a time sequence recording mode. The two types of data are mutually independent in the processes of acquisition, storage and analysis, and lack of effective technical means to realize the mining and fusion of the internal association, so that the cognitive ability of the management system stays at the surface layer statistics level. A core drawback of the prior art solution is its stationarity and priors of the analysis dimensions. Because of the lack of an effective method for deep fusion and pattern analysis of multi-source heterogeneous data, the construction of the current management framework is seriously dependent on the manual experience of field experts, and the limited and fixed management dimensions such as equipment types, physical positions, brand models and the like are preset. The dimension definition mode based on priori knowledge cannot automatically identify and extract potential and dynamic evolution association modes from complex interaction data generated in the actual operation process of equipment. The management view is thus solidified, and the system is difficult to adapt to new rules and new relations which are continuously emerging in the equipment use scene. This technical limitation results in hysteresis and one-sidedness of the management decisions. In the face of increasingly networked and intelligent sports fitness facilities, the traditional method cannot provide operators with dynamic and multidimensional management insight driven by data, so that the realization of advanced management targets such as resource optimal allocation, preventive maintenance, personalized service and the like is restricted. There is a need for a solution that automatically discovers and constructs a characterization management dimension from the fused data. Disclosure of Invention The invention aims to provide a physical exercise equipment management method based on an artificial intelligence algorithm, so as to solve the problems in the background technology. To achieve the above object, the present invention provides a sports fitness equipment management method based on an artificial intelligence algorithm, the method comprising: fusing static description text and dynamic use log of the sports fitness equipment to generate a mixed information stream; Performing mode slicing on the mixed information flow, and separating structural characteristic slices and functional characteristic slices of equipment; excavating a static topological relation network of equipment from the structural feature slice, and capturing a dynamic behavior track sequence of the equipment from the functional feature slice; performing space-time alignment operation on the static topological relation network and the dynamic behavior track sequence to form a joint characterization vector of equipment; Constructing a relationship graph model taking the equipment as a center by utilizing the joint characterization vector of the equipment, wherein nodes of the relationship graph model comprise equipment entities, and edges comprise static topological relations and dynamic behavior associations; and traversing all edges in the relational graph model, identifying and aggregating edge clusters with the same association mode, and abstracting each edge cluster into a management dimension to obtain a multidimensional management framework. Preferably, the static description text includes specification parameters, material composition and installation configuration information of equipment, the dynamic usage log includes running time record, usage frequency record, maintenance operation record and fault alarm record of the equipment, the mixed information stream is a continuous data record formed by interweaving and splicing the static description text and the dynamic usage log with time stamps or logical association points, the structural feature slice is a segment set extracted from the mixed information stream and used for describing physical composition and connection relation of the equipment, the functional feature slice is a segment set extracted from the mixed information stream and used for describing running state and interaction behavior of the equipment, the static topological relation netwo