Knowledge graph optimization method and system for large traffic models

Abstract

A knowledge graph optimization method and system for large traffic models is provided. The method comprises monitoring and filtering traffic data processing tasks of large traffic models, such as neur

Description

The present invention relates to the field of large traffic models, and in particular to a knowledge graph optimization method and a system for large traffic models. Background Technology Rapid changes in urban road networks and car ownership have made urban traffic conditions increasingly complex. In order to accurately and continuously predict urban traffic conditions, especially traffic flows, corresponding large traffic models will be constructed based on existing urban traffic data, and the large traffic models will be used for calculating and predicting urban traffic conditions. When large traffic models are used for prediction, different data processing tasks or different traffic prediction stages will yield corresponding data results, which effectively reflect future changes in urban traffic over time periods. However, due to the extensive nature and singular format of the data results, users are unable to intuitively and accurately extract the information they need, thus failing to provide reliable data support for subsequent urban traffic planning. Additionally, the data results from the large traffic models cannot be graphically transformed, reducing flexibility of using the large traffic models in different scenarios. Summary of the Invention The purposes of the present invention is to provide a knowledge graph optimization method and a system for large traffic models, and in the present invention, traffic data processing tasks of large traffic models are monitored to obtain traffic status data of key traffic nodes which are clustered to obtain traffic feature data of the key traffic nodes, thus providing reliable data for constructing knowledge graphs; based on the traffic feature data, the present invention generate graph layers under the knowledge graphs corresponding to the key traffic nodes, and integrates the knowledge graphs of the key traffic nodes into global knowledge graphs based on associated relationships of the key traffic nodes within traffic maps; additionally, the present invention updates the global knowledge graphs based on status output from traffic prediction results of the large traffic models to generate knowledge graph sets, sorts out predictive outputs of the large traffic models at different stages in a form of knowledge graphs, retrieves matched global knowledge graphs from the knowledge graph sets and transmit the same to user terminals for visual display, and provides the user terminals with intuitive and accurate traffic prediction knowledge graph information, thereby providing flexibility of using the large traffic models in different scenarios. The present invention is implemented through following technical solutions: A knowledge graph optimization method for large traffic models comprises: monitoring and filtering traffic data processing tasks of large traffic models to obtain traffic status data of key traffic nodes, and clustering the traffic status data to obtain traffic feature data of the key traffic nodes; generating graph layers under knowledge graphs corresponding to the key traffic nodes based on the traffic feature data, and conducting associated identification processing for the graph layers under the knowledge graphs; and integrating knowledge graphs of the key traffic nodes into global knowledge graphs of the key traffic nodes based on associated relationships of the key traffic nodes within traffic maps corresponding to the large traffic models; and updating the global knowledge graphs based on status output from traffic prediction results of the large traffic models, so as to generate knowledge graph sets correspondingly; and based on knowledge graph retrieval requests from user terminals, retrieving matched global knowledge graphs from the knowledge graph sets and transmitting the same to the user terminals for visual display. Preferably, the steps of monitoring and filtering traffic data processing tasks of large traffic models to obtain traffic status data of key traffic nodes, and clustering the traffic status data to obtain traffic feature data of the key traffic nodes comprise: based on model network layers where traffic data processing tasks of the large traffic models are located, monitoring entire processes of the traffic data processing tasks to obtain traffic status change data corresponding to the key traffic nodes within target traffic areas when the traffic data processing tasks are executed; and identifying and removing abnormal data from the traffic status change data to obtain traffic status data of the key traffic nodes; and identifying data types of the traffic status data to obtain data types corresponding to all sub-data under the traffic status data; and based on the data types, clustering all sub-data under the traffic status data to obtain traffic feature data of the key traffic nodes, wherein the traffic feature data comprises road network connection feature data, traffic flow feature data and traffic facility feature data of the key traffic nodes. Preferably the steps of generating graph layers under knowledge graphs corresponding to the key traffic nodes based on the traffic feature data, and conducting associated identification processing for the graph layers under the knowledge graphs; integrating knowledge graphs of the key traffic nodes into global knowledge graphs of the key traffic nodes based on associated relationships of the key traffic nodes within traffic maps corresponding to the large traffic models comprise: performing metadata identification on the road network connection feature data, the traffic flow feature data, and the traffic facility feature data comprised in the traffic feature data, respectively, to obtain road network connection metadata sets, traffic flow metadata sets, and traffic facility metadata sets correspondingly; and based on the road network connection metadata sets, the traffic flow metadata sets, and the traffic facility metadata sets, respectively constructing road network graph layers, traffic flow graph layers, and traffic facility graph layers under the knowledge graphs corresponding to the key traffic nodes, and conducting associated identification processing for the road network graph layers, the traffic flow graph layers, and the traffic facility graph layers of the key traffic nodes; and based on positional relationships of the key traffic nodes within the traffic maps corresponding to the large traffic models, determining relative positions and relative distances of the key traffic nodes; and based on the relative positions and the relative distances, integrating the knowledge graphs of the key traffic nodes to obtain global knowledge graphs of the key traffic nodes. Preferably, the steps of updating the global knowledge graphs based on status output from traffic prediction results of the large traffic models, so as to generate knowledge graph sets correspondingly; and based on knowledge graph retrieval requests from user terminals, retrieving matched global knowledge graphs from the knowledge graph sets and transmitting the same to the user terminals for visual display comprise: obtaining traffic prediction results of the large traffic models at preset time intervals during traffic prediction processes, identifying the traffic prediction results, and determining traffic prediction data of the key traffic nodes comprised in the traffic prediction results; and based on the traffic prediction data, updating knowledge graph portions of the key traffic nodes corresponding to the global knowledge graphs, so as to generate knowledge graph sets correspondingly, wherein updated global knowledge graphs comprised in the knowledge graph sets correspond to the traffic prediction results one by one ; and parsing knowledge graph retrieval requests from user terminals and determining key traffic node attribute information to which knowledge graphs expected by the user terminals belong; and based on the key traffic node attribute information, retrieving matched global knowledge graphs from the knowledge graph sets, packaging retrieved matched global knowledge graphs and then transmitting the same to the user terminals for visual display. Preferably, the step of packaging retrieved matched global knowledge graphs and then transmitting the same to the user terminals for visual display comprises: SI: using a following formula (1), based on displayed widths and heights of the matched global knowledge graphs and displayed widths and heights of the user terminals, to determine initial scaling ratios of the matched global knowledge graphs, If = y where, represents initial scaling ratios of the matched global knowledge graphs, represents displayed length values of the user terminals, - represents D„ displayed width values of the user terminals, represents displayed length L values of the matched global knowledge graphs, ‘ represents displayed width values of the matched global knowledge graphs, represents taking minimum values on both sides of a comma enclosed in parentheses; S2: using a following formula (2), based on display ratios of the user terminals and total numbers of pixels at maximum displayable resolutions, as well as display ratios of the matched global knowledge graphs and total numbers of pixels at current resolutions, to determine numbers of pixels to be removed from rows and columns of the matched global knowledge graphs: AM where represents numbers of vertical columns to be removed from rows and columns of the matched global knowledge graphs; ' 1 represents numbers of horizontal columns to be removed from rows and columns of the matched global X knowledge graphs; "represents display ratios of the matched global knowledge _ •1 C P y _ V — Oux / V — graphs; "represents display ratios of the user terminals; „ m y = = m a? indicates conditions where and ' hold true; 1 and “ represent integer values of aspect ratios in display ratios of the matched global 11. knowledge graphs; and represent integer values of aspect ratios in display ratios of the user terminals; represents taking maximum values on both sides of a comma enclosed in brackets; and S3: using a following formula (3), based on numbers of pixels to be removed from rows and columns of the matched global knowledge graphs and displayed widths and heights of the user terminals, to determine final magnification ratios for displaying the matched global knowledge graphs, where, n represents final magnification ratios for displaying the matched global a knowledge graphs; represents vertical width values of single pixel blocks of the matched global knowledge graphs; and ' horizontal width values of single pixel blocks of the matched global knowledge graphs. A knowledge graph optimization system for large traffic models comprises: a traffic status data acquisition module, used for monitoring and filtering traffic data processing tasks of large traffic models, and obtaining traffic status data of key traffic nodes; a traffic feature data generation module, used for clustering the traffic status data and obtaining traffic feature data of the key traffic nodes; a knowledge graph generation module, used for generating graph layers under knowledge graphs corresponding to the key traffic nodes based on the traffic feature data, and performing associated identification processing on the graph layers under the knowledge graphs; a knowledge graph integration module, used for integrating the knowledge graphs of the key traffic nodes based on associated relationships of the key traffic nodes within traffic maps corresponding to the large traffic models, and obtaining global knowledge graphs of the key traffic nodes; a knowledge graph update module, used for updating the global knowledge graphs based on status output from traffic prediction results of the large traffic models, and generating knowledge graph sets correspondingly; and a knowledge graph retrieval and transmission module, used for retrieving matched global knowledge graphs from the knowledge graph sets based on knowledge graph retrieval requests from user terminals and transmitting the same to the user terminals for visual display. Preferably, the traffic status data acquisition module, used for monitoring and filtering traffic data processing tasks of large traffic models, and obtaining traffic status data of key traffic nodes, comprises: based on model network layers where traffic data processing tasks of large traffic models are located, monitoring and obtaining traffic status change data corresponding to the key traffic nodes in target traffic areas when the traffic data processing tasks are executed, identifying and eliminating abnormal data from the traffic status change data, and obtaining traffic status data of the key traffic nodes; and the traffic feature data generation module, used for clustering the traffic status data and obtaining traffic feature data of the key traffic nodes, comprises: identifying data types of the traffic status data to obtain data types corresponding to all sub-data under the traffic status data; based on the data types, clustering all sub-data under the traffic status data to obtain traffic feature data of the key traffic nodes, wherein the traffic feature data comprises road network connection feature data, traffic flow feature data and traffic facility feature data of the key traffic nodes. Preferably, the knowledge graph generation module, used for generating several graph layers under the knowledge graphs corresponding to the key traffic nodes based on the traffic feature data, and performing associated identification processing on all graph layers under the knowledge graphs, comprises: performing metadata identification on road network connection feature data, traffic flow feature data and traffic facility feature data comprised in the traffic feature data to obtain corresponding road network connection metadata sets, traffic flow metadata sets and traffic facility metadata sets; based on the road network connection metadata sets, the traffic flow metadata sets and the traffic facility metadata sets, respectively constructing road network graph layers, traffic flow graph layers and the traffic facility graph layers under knowledge graphs corresponding to the key traffic nodes, and the road network map layer, performing associated identification processing of the key traffic nodes on the road network graph layers, the traffic flow graph layers and the traffic facility graph layers; the knowledge graph integration module, used for integrating the knowledge graphs of the key traffic nodes based on associated relationships of the key traffic nodes within traffic maps corresponding to the large traffic models, and obtaining global knowledge graphs of the key traffic nodes, comprises : based on positional relationships of the key traffic nodes within the traffic maps corresponding to the large traffic models, determining relative positions and relative distances of the key traffic nodes; then based on the relative positions and the relative distances, integrating knowledge graphs of the key traffic nodes to obtain global knowledge maps of the key traffic nodes. Preferably the knowledge graph update module, used for updating the global knowledge graphs based on status output from traffic prediction results of the large traffic models, and generating knowledge graph sets correspondingly, comprises: retrieving all traffic prediction results of large traffic models at each preset time intervals during traffic prediction processes, identifying all traffic prediction results, and determining traffic prediction data of the key traffic nodes comprised in traffic prediction results; based on the traffic prediction data, updating knowledge graph portions of the key traffic nodes corresponding to the global knowledge graphs, and generating knowledge graph sets correspondingly; wherein all updated knowledge graphs comprised in the knowledge graph sets correspond to all traffic prediction results one by one; and the knowledge graph retrieval and transmission module, used for retrieving matched global knowledge graphs from the knowledge graph sets based on knowledge graph retrieval requests from user terminals and transmitting the same to the user terminals for visual display, comprises: parsing the knowledge graph retrieval requests from user terminals and determining key traffic node attribute information of the knowledge graphs that the user terminals expect to obtain; and based on the key traffic node attribute information, retrieving matched global knowledge graphs from the knowledge graph sets, and then packaging matched global knowledge graphs and transmitting the same to the user terminals for visual display. Compared to existing technology, the present invention has following beneficial effects: The present invention provides a knowledge graph optimization method and a system for large traffic models, so as to monitor traffic data processing tasks of large traffic models, obtain traffic status data of key traffic nodes, and cluster and process the same to obtain traffic feature data of the key traffic nodes, thus providing reliable data for constructing knowledge graphs. Furthermore, based on the traffic feature data, the present invention generates graph layers under the knowledge graphs corresponding to the key traffic nodes, and integrates the knowledge graphs of the key traffic nodes into global knowledge graphs based on associated relationships of the key traffic nodes within traffic maps. Additionally, the present invention updates the global knowledge graphs based on status output from traffic prediction results of the large traffic models to generate knowledge graph sets, organizes predictive outputs of the large traffic models at different stages in a form of knowledge graphs and retrieves matched global knowledge graphs from collected knowledge graphs to transfer the same to user terminals for visual display, and provides users with intuitive and accurate traffic prediction knowledge graph information, thereby enhances flexibility of using large traffic models in various scenarios. Brief Description of the Drawings In order to more clearly illustrate embodiments of the present invention or technical solutions in the prior art, drawings required for use in the embodiments or description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work. Figure 1 is a flow chart of a knowledge graph optimization method for large traffic models provided by the present invention. Figure 2 is a structural diagram of a knowledge graph optimization system for large traffic models provided by the present invention. Specific Embodiments In order to make above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings. It is to be understood that specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only some structures related to the present invention are shown in the accompanying drawings, rather than all structures. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention. The terms "comprises" and "comprising" and any variations thereof in the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but may optionally comprise steps or units not listed, or may optionally comprise other steps or units inherent to these processes, methods, products or devices. Reference to "embodiments" herein means that a particular feature, structure, or feature described in conjunction with the embodiments may be comprised in at least one embodiment of the present invention. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments. Please refer to figure 1, an embodiment of the present invention provides a knowledge graph optimization method for large traffic models, comprising: monitoring and filtering traffic data processing tasks of large traffic models to obtain traffic status data of key traffic nodes, and clustering the traffic status data to obtain traffic feature data of the key traffic nodes; generating graph layers under knowledge graphs corresponding to the key traffic nodes based on the traffic feature data, and conducting associated identification processing for the graph layers under the knowledge graphs; integrating knowledge graphs of the key traffic nodes into global knowledge graphs of the key traffic nodes based on associated relationships of the key traffic nodes within traffic maps corresponding to the large traffic models; and updating the global knowledge graphs based on status output from traffic prediction results of the large traffic models, so as to generate knowledge graph sets correspondingly; and based on knowledge graph retrieval requests from user terminals, retrieving matched global knowledge graphs from the knowledge graph sets and transmitting the same to the user terminals for visual display. The above embodiment has following beneficial effects: the knowledge graph optimization method for large traffic models monitors traffic data processing tasks of large traffic models, obtains the traffic status data of key traffic nodes, and clusters the same to obtain traffic feature data of the key traffic nodes, thereby providing reliable data for constructing knowledge graphs; based on the traffic feature data, graph layers under the knowledge graphs corresponding to the key traffic nodes are generated, and based on associated relationships of the key traffic nodes within traffic maps, the knowledge graphs of the key traffic nodes are integrated into global knowledge graphs, and overall prediction outputs of the large traffic models are represented in global graphs; based on status output from the traffic prediction results of the large traffic models, the global knowledge graphs are updated to generate knowledge graph sets, prediction outputs of the large traffic models at different stages are organized in the form of knowledge graphs, and matched global knowledge graphs are retrieved from the knowledge graph sets and transmitted to user terminals for visual display, thereby providing the user terminals with intuitive and accurate traffic prediction knowledge graph information, and improving the flexibility of using the large traffic models in different scenarios. In another embodiment, the "monitoring and filtering traffic data processing tasks of large traffic models to obtain traffic status data of key traffic nodes, and clustering the traffic status data to obtain traffic feature data of the key traffic nodes" comprises: based on model network layers where traffic data processing tasks of the large traffic models are located, monitoring entire processes of the traffic data processing tasks to obtain traffic status change data corresponding to the key traffic nodes within target traffic areas when the traffic data processing tasks are executed; and identifying and removing abnormal data from the traffic status change data to obtain traffic status data of the key traffic nodes; and identifying data types of the traffic status data to obtain data types corresponding to all sub-data under the traffic status data; and based on the data types, clustering all sub-data under the traffic status data to obtain traffic feature data of the key traffic nodes, wherein the traffic feature data comprises road network connection feature data, traffic flow feature data and traffic facility feature data of the key traffic nodes. The above embodiment has following beneficial effects: the large traffic models are neural network models, and accordingly, the large traffic models comprise model network layers, when the large traffic models execute traffic data processing tasks, all model network layers will perform corresponding traffic data processing operations, and when the large traffic models execute the traffic data processing task, corresponding model network layers are monitored throughout entire processes, and traffic status change data corresponding to the key traffic nodes in target traffic areas during the execution of the traffic data processing tasks can be obtained, wherein the key traffic nodes can be, but not limited to, traffic location points such as crossroads, tunnels, bridges, etc. in the target traffic areas, and the traffic status change data can be, but not limited to, traffic flows, pedestrian flows and physical environment change data of the key traffic nodes; then, abnormal data such as the erroneous data components in the traffic status change data is identified and removed, so as to obtain the traffic status data of the key traffic nodes, thereby ensuring the reliability of the traffic status data; further, the traffic status data is also identified as data types, and all sub-data under the traffic status data are clustered based on the data types, so as to obtain the traffic feature data of the key traffic nodes, so that the traffic feature data can comprehensively characterize traffic features of corresponding dimensions. In another embodiment of the present invention, the "generating graph layers under knowledge graphs corresponding to the key traffic nodes based on the traffic feature data, and conducting associated identification processing for the graph layers under the knowledge graphs; integrating knowledge graphs of the key traffic nodes into global knowledge graphs of the key traffic nodes based on associated relationships of the key traffic nodes within traffic maps corresponding to the large traffic models" comprises: performing metadata identification on the road network connection feature data, the traffic flow feature data, and the traffic facility feature data comprised in the traffic feature data, respectively, to obtain road network connection metadata sets, traffic flow metadata sets, and traffic facility metadata sets correspondingly; and based on the road network connection metadata sets, the traffic flow metadata sets, and the traffic facility metadata sets, respectively constructing road network graph layers, traffic flow graph layers, and traffic facility graph layers under the knowledge graphs corresponding to the key traffic nodes, and conducting associated identification processing for the road network graph layers, the traffic flow graph layers, and the traffic facility graph layers of the key traffic nodes; and based on positional relationships of the key traffic nodes within the traffic maps corresponding to the large traffic models, determining relative positions and relative distances of the key traffic nodes; and based on the relative positions and the relative distances, integrating the knowledge graphs of the key traffic nodes to obtain global knowledge graphs of the key traffic nodes. The above embodiment has following beneficial effects: since the amount of data corresponding to road network connection feature data, traffic flow feature data and traffic facility feature data comprised in the traffic feature data is large, if above three kinds of feature data are directly converted into knowledge graphs, data redundancy will occur in the knowledge graphs; to this end, metadata identification is performed on road network connection feature data, traffic flow feature data and traffic facility feature data comprised in the traffic feature data, respectively, to obtain corresponding road network connection metadata sets, traffic flow metadata sets and traffic facility metadata sets, wherein the metadata identification may be but is not limited to data keyword recognition, so that it can be ensured that the metadata obtained by recognition fully reflects semantic contents of corresponding feature data. Based on the road network connection metadata sets, the traffic flow metadata sets and the traffic facility metadata sets, road network graph layers, traffic flow graph layers and traffic facility graph layers under the knowledge graphs corresponding to the key traffic nodes are respectively constructed, and the road network graph layers, the traffic flow graph layers and the traffic facility graph layers are processed for associated identification of the key traffic nodes to which they belong, so that the knowledge graphs can be constructed in a hierarchical manner, the knowledge graphs can fully comprise corresponding traffic feature data, and data structures of the knowledge graphs can also be optimized; different key traffic nodes correspond to different location points within the target traffic areas, accordingly, different key traffic nodes have corresponding positional correlations on the traffic maps, therefore, based on positional relationships of the key traffic nodes within the traffic map corresponding to the large traffic models, the relative positions and relative distances of the key traffic nodes are determined, and the knowledge graphs of the key traffic nodes are integrated to obtain global knowledge graphs about the key traffic nodes; in this way, the knowledge graphs of the key traffic nodes can be organically integrated while maintaining the independence of the knowledge graphs of the key traffic nodes, thereby avoiding crosstalk between the knowledge graphs of different key traffic nodes. In another embodiment, the "updating the global knowledge graphs based on status output from traffic prediction results of the large traffic models, so as to generate knowledge graph sets correspondingly; and based on knowledge graph retrieval requests from user terminals, retrieving matched global knowledge graphs from the knowledge graph sets and transmitting the same to the user terminals for visual display" comprises: obtaining traffic prediction results of the large traffic models at preset time intervals during traffic prediction processes, identifying the traffic prediction results, and determining traffic prediction data of the key traffic nodes comprised in the traffic prediction results; based on the traffic prediction data, updating knowledge graph portions of the key traffic nodes corresponding to the global knowledge graphs, so as to generate knowledge graph sets correspondingly, wherein updated global knowledge graphs comprised in the knowledge graph sets correspond to the traffic prediction results one by one ; and parsing knowledge graph retrieval requests from user terminals and determining key traffic node attribute information to which knowledge graphs expected by the user terminals belong; and based on the key traffic node attribute information, retrieving matched global knowledge graphs from the knowledge graph sets, packaging retrieved matched global knowledge graphs and then transmitting the same to the user terminals for visual display. The above embodiment has following beneficial effects: large traffic models output corresponding traffic prediction results at every preset time interval when traffic prediction (such as traffic flow prediction) is conducted, and traffic prediction results correspond to different key traffic nodes, for this purpose, all traffic prediction results are identified, and traffic prediction data of the key traffic nodes comprised in each traffic prediction result is determined, that is, it is determined which key traffic node each traffic prediction result corresponds to, and the traffic prediction data comprised in the traffic prediction results is extracted, so as to update the knowledge graph portions of the key traffic nodes corresponding to the global knowledge graphs, and each time an update is completed, updated global knowledge graphs are obtained, and then all updated global knowledge graphs are formed into knowledge graph sets, so as to perform knowledge graph conversion on the prediction results generated by the large traffic models at different prediction stages in the traffic prediction processes; then knowledge graph retrieval requests of user terminals are parsed and processed, and key traffic node attribute information of the knowledge graphs to which the user terminals expect to obtain belongs is determined, so as to retrieve matched global knowledge graphs from the knowledge graph sets, and then retrieved matched global knowledge graphs are packaged and transmitted to the user terminals for visual display, so as to optimize the display intuitiveness and reliability of the knowledge graphs corresponding to the large traffic models. In another embodiment, the "packaging retrieved matched global knowledge graphs and then transmitting the same to the user terminals for visual display" comprises: SI: using a following formula (1), based on displayed widths and heights of the matched global knowledge graphs and displayed widths and heights of the user terminals, to determine initial scaling ratios of the matched global knowledge graphs, K = mm(Dv / Dc,. I )   (1) where, represents initial scaling ratios of the matched global knowledge graphs, D T V                                                                                                                           f j represents displayed length values of the user terminals, y represents displayed width values of the user terminals, 5 represents displayed length £ $ values of the matched global knowledge graphs, represents displayed width values of the matched global knowledge graphs, represents taking minimum values on both sides of a comma enclosed in parentheses; S2: using a following formula (2), based on display ratios of the user terminals and total numbers of pixels at maximum displayable resolutions, as well as display ratios of the matched global knowledge graphs and total numbers of pixels at current resolutions, to determine numbers of pixels to be removed from rows and columns of the matched global knowledge graphs: A where represents numbers of vertical columns to be removed from rows and AiV columns of the matched global knowledge graphs; represents numbers of horizontal columns to be removed from rows and columns of the matched global knowledge graphs; represents display ratios of the matched global knowledge graphs; "represents display ratios of the user terminals; A,= — y „ m a* indicates conditions where A and ' hold true; 1 and A represent integer values of aspect ratios in display ratios of the matched global •m ?! knowledge graphs; and represent integer values of aspect ratios in display ratios of the user terminals; M represents taking maximum values on both sides of a comma enclosed in brackets; S3: using a following formula (3), determining final magnification ratios for displaying the matched global knowledge graphs based on numbers of pixels to be removed from rows and columns of the matched global knowledge graphs and displayed widths and heights of the user terminals, to determine final magnification ratios for displaying the matched global knowledge graphs, where In the formula above (3), represents final magnification ratios for displaying the matched global knowledge graphs; represents vertical width values of single pixel blocks of the matched global knowledge graphs; horizontal width values of single pixel blocks of the matched global knowledge graphs. The above embodiment has following beneficial effects: initial scaling ratios of the matched global knowledge graphs are determined according to displayed lengths and widths of the matched global knowledge graphs and displayed lengths and widths of the user terminals by using the above formula (1), so that entire images of the matched global knowledge graphs are first scaled to display interfaces of the user terminals, which is convenient for subsequent fine-tuning and improves the efficiency of the system; then, numbers of pixels to be removed from rows and columns of the matched global knowledge graphs are determined according to the display ratios of the user terminals and total numbers of pixels at maximum displayable resolutions, as well as the display ratios of the matched global knowledge graphs and total numbers of pixels at current resolutions by using the above formula (2), so that length and width ratios are made suitable for the display interfaces of the user terminals by reducing the pixels as much as possible, so that users can watch more comfortably; finally, the final magnification ratios for displaying the matched global knowledge graphs are determined according to the numbers of pixels to be removed from the rows and columns of the matched global knowledge graphs and the displayed lengths and widths of the user terminals by using the above formula (3), so that the matched global knowledge graphs can fill the display interfaces of the user terminals as much as possible, thereby ensuring the convenience of user observation. Please refer to figure 2, an embodiment of the present invention provides a knowledge graph optimization system for large traffic models comprising: a traffic status data acquisition module, used for monitoring and filtering traffic data processing tasks of large traffic models, and obtaining traffic status data of key traffic nodes; a traffic feature data generation module , used for clustering the traffic status data and obtaining traffic feature data of the key traffic nodes; a knowledge graph generation module, used for generating graph layers under knowledge graphs corresponding to the key traffic nodes based on the traffic feature data, and performing associated identification processing on the graph layers under the knowledge graphs; a knowledge graph generation module, used for integrating the knowledge graphs of the key traffic nodes based on associated relationships of the key traffic nodes within traffic maps corresponding to the large traffic models, and obtaining global knowledge graphs of the key traffic nodes; a knowledge graph integration module, used for integrating the knowledge graphs of the key traffic nodes based on the association relationship of key traffic nodes in the traffic map corresponding to the traffic model, and obtain the global knowledge graph of key traffic nodes; a knowledge graph update module, used for updating the global knowledge graphs based on status output from traffic prediction results of the large traffic models, and generating knowledge graph sets correspondingly; and a knowledge graph retrieval and transmission module, used for retrieving matched global knowledge graphs from the knowledge graph sets based on knowledge graph retrieval requests from user terminals and transmitting the same to the user terminals for visual display. The above embodiment has following beneficial effects: the knowledge graph optimization system for large traffic models monitors traffic data processing tasks of large traffic models, obtains the traffic status data of key traffic nodes, and clusters the same to obtain traffic feature data of the key traffic nodes, providing reliable data for constructing knowledge graphs; based on the traffic feature data, graph layers under the knowledge graphs corresponding to the key traffic nodes are generated, and based on associated relationships of the key traffic nodes within traffic maps, the knowledge graphs of the key traffic nodes are integrated into global knowledge graphs, and overall prediction outputs of the large traffic models are represented in global graphs; based on status output from traffic prediction results of the large traffic models, the global knowledge graphs are updated to generate knowledge graph sets, prediction outputs of the large traffic models at different stages are organized in the form of knowledge graphs, and matched global knowledge graphs is retrieved from the knowledge graph sets and transmitted to user terminals for visual display, thereby providing the user terminals with intuitive and accurate traffic prediction knowledge graph information, and improving the flexibility of using the large traffic models in different scenarios. In another embodiment, the traffic status data acquisition module, used for monitoring and filtering traffic data processing tasks of large traffic models, and obtaining traffic status data of key traffic nodes, comprises: based on the model network layers where the traffic data processing tasks of the traffic model are located, monitoring and obtaining traffic status change data corresponding to the key traffic nodes in target traffic areas when the traffic data processing tasks are executed; identifying and eliminating abnormal data from the traffic status change data, and obtaining traffic status data of the key traffic nodes; and the traffic feature data generation module, used for clustering the traffic status data and obtaining traffic feature data of the key traffic nodes, comprises: identifying data types of the traffic status data to obtain data types corresponding to all sub-data under the traffic status data; based on the data types, clustering all sub-data under the traffic status data to obtain traffic feature data of the key traffic nodes; wherein the traffic feature data comprises road network connection feature data, traffic flow feature data and traffic facility feature data of the key traffic nodes. The above embodiment has following beneficial effects: the large traffic models are neural network models, and accordingly, the large traffic models comprise model network layers, when the large traffic models execute traffic data processing tasks, all model network layers will perform corresponding traffic data processing operations, and when the large traffic models execute the traffic data processing task, corresponding model network layers are monitored throughout entire processes, and traffic status change data corresponding to the key traffic nodes in target traffic areas during the execution of the traffic data processing tasks can be obtained, wherein the key traffic nodes can be, but not limited to, traffic location points such as crossroads, tunnels, bridges, etc. in the target traffic areas, and the traffic status change data can be, but not limited to, traffic flows, pedestrian flows and physical environment change data of the key traffic nodes; then, abnormal data such as the erroneous data components in the traffic status change data is identified and removed, so as to obtain the traffic status data of the key traffic nodes, thereby ensuring the reliability of the traffic status data; further, the traffic status data is also identified as data types, and all sub-data under the traffic status data are clustered based on the data types, so as to obtain the traffic feature data of the key traffic nodes, so that the traffic feature data can comprehensively characterize traffic features of corresponding dimensions. In another embodiment, the knowledge graph generation module, used for generating graph layers under the knowledge graphs corresponding to the key traffic nodes based on the traffic feature data, and performing associated identification processing on the graph layers under the knowledge graphs, comprises: performing metadata identification on road network connection feature data, traffic flow feature data and traffic facility feature data comprised in the traffic feature data to obtain corresponding road network connection metadata sets, traffic flow metadata sets and traffic facility metadata sets; based on the road network connection metadata sets, the traffic flow metadata sets and the traffic facility metadata sets, respectively constructing road network graph layers, traffic flow graph layers and the traffic facility graph layers under knowledge graphs corresponding to the key traffic nodes, and the road network map layer, and performing associated identification processing of the key traffic nodes on the road network graph layers, the traffic flow graph layers and the traffic facility graph layers; and the knowledge graph integration module, used for integrating knowledge graphs of the key traffic nodes based on associated relationships of the key traffic nodes within traffic maps corresponding to large traffic models, and obtaining global knowledge graphs about key traffic nodes, comprises : based on positional relationships of the key traffic nodes within the traffic maps corresponding to the large traffic models, determining relative positions and relative distances of the key traffic nodes; then based on the relative positions and the relative distances, integrating knowledge graphs of the key traffic nodes to obtain global knowledge maps of the key traffic nodes. The above embodiment has following beneficial effects: since the amount of data corresponding to road network connection feature data, traffic flow feature data and traffic facility feature data comprised in the traffic feature data is large, if the above three kinds of feature data are directly converted into knowledge graphs, data redundancy will occur in the knowledge graphs; to this end, metadata identification is performed on road network connection feature data, traffic flow feature data and traffic facility feature data comprised in the traffic feature data, respectively, to obtain corresponding road network connection metadata sets, traffic flow metadata sets and traffic facility metadata sets, wherein the metadata identification may be but is not limited to data keyword recognition, so that it can be ensured that the metadata obtained by recognition fully reflects semantic contents of corresponding feature data. Based on the road network connection metadata sets, the traffic flow metadata sets and the traffic facility metadata sets, road network graph layers, traffic flow graph layers and traffic facility graph layers under the knowledge graphs corresponding to the key traffic nodes are respectively constructed, and the road network graph layers, the traffic flow graph layers and the traffic facility graph layers are processed for associated identification of the key traffic nodes to which they belong, so that the knowledge graphs can be constructed in a hierarchical manner, the knowledge graphs can fully comprise corresponding traffic feature data, and data structures of the knowledge graphs can also be optimized; different key traffic nodes correspond to different location points within the target traffic areas, accordingly, different key traffic nodes have corresponding positional correlations on the traffic maps, therefore, based on positional relationships of the key traffic nodes within the traffic map corresponding to the large traffic models, the relative positions and relative distances of the key traffic nodes are determined, and the knowledge graphs of the key traffic nodes are integrated to obtain global knowledge graphs about the key traffic nodes; in this way, the knowledge graphs of the key traffic nodes can be organically integrated while maintaining the independence of the knowledge graphs of the key traffic nodes, thereby avoiding crosstalk between the knowledge graphs of different key traffic nodes. In another embodiment, the knowledge graph update module, used for updating the global knowledge graphs based on output status of traffic prediction results of the large traffic models, and generating corresponding knowledge graph sets, comprises: retrieving all traffic prediction results of large traffic models at each preset time intervals during traffic prediction processes, identifying all traffic prediction results, and determining traffic prediction data of the key traffic nodes comprised in traffic prediction results; based on the traffic prediction data, updating knowledge graph portions of the key traffic nodes corresponding to the global knowledge graphs, and generating corresponding knowledge graph sets; wherein all updated knowledge graphs comprised in the knowledge graph sets correspond to all traffic prediction results one by one; and the knowledge graph retrieval and transmission module, used for retrieving matched global knowledge graphs from the knowledge graph sets based on knowledge graph acquisition requests from the user terminals and transmitting the same to the user terminals for visual display, comprises: parsing the knowledge graph acquisition requests from the user terminals and determining key traffic node attribute information of the knowledge graphs that the user terminals expect to obtain; and based on the key traffic node attribute information, retrieving matched global knowledge graphs from the knowledge graph sets, and then packaging matched global knowledge graphs and transmitting the same to the user terminals for visual display. The above embodiment has following beneficial effects: large traffic models output corresponding traffic prediction results at every preset time interval when the traffic prediction (such as traffic flow prediction) is conducted, and traffic prediction results correspond to different key traffic nodes, for this purpose, all traffic prediction results are identified, and traffic prediction data of the key traffic nodes comprised in each traffic prediction result is determined, that is, it is determined which key traffic node each traffic prediction result corresponds to, and the traffic prediction data comprised in the traffic prediction result is extracted, so as to update the knowledge graph portions of the key traffic nodes corresponding to the global knowledge graphs, and each time an update is completed, updated global knowledge graphs are obtained, and then all updated global knowledge graphs are formed into knowledge graph sets, so as to perform knowledge graph conversion on the prediction results generated by the large traffic models at different prediction stages in the traffic prediction processes; then knowledge graph retrieval requests of user terminals are parsed and processed, and key traffic node attribute information of the knowledge graphs to which the user terminals expect to obtain belongs is determined, so as to retrieve matched global knowledge graphs from the knowledge graph sets, and then retrieved matched global knowledge graphs are packaged and transmitted to the user terminals for visual display, so as to optimize the display intuitiveness and reliability of the knowledge graphs corresponding to the large traffic models. To sum up, in the knowledge graph optimization method and the system for large traffic models of the present invention, traffic data processing tasks of large traffic models are monitored to obtain traffic status data of key traffic nodes which are clustered and processed to obtain traffic feature data of the key traffic nodes, thus providing reliable data for constructing knowledge graphs; based on the traffic feature data, the present invention generate graph layers under the knowledge graphs corresponding to the key traffic nodes, and integrates the knowledge graphs of the key traffic nodes into global knowledge graphs based on associated relationships of the key traffic nodes within traffic maps; additionally, the present invention updates the global knowledge graphs based on status output from traffic prediction results of the large traffic models to generate knowledge graph sets, sorts out predictive outputs of the large traffic models at different stages in a form of knowledge graphs, retrieves matched global knowledge graphs from the knowledge graph sets and transmit the same to user terminals for visual display, and provides the user terminals with intuitive and accurate traffic prediction knowledge graph information. The above are only specific embodiments of the present invention, and any other improvements made based on the concept of the present invention are considered to be within the protection scope of the present invention.

Claims

1. A knowledge graph optimization method for large traffic models, comprising:monitoring and filtering traffic data processing tasks of large traffic models to obtain traffic status data of key traffic nodes, and clustering the traffic status data to obtain traffic feature data of the key traffic nodes;generating graph layers under knowledge graphs corresponding to the key traffic nodes based on the traffic feature data, and conducting associated identification processing for the graph layers under the knowledge graphs; and integrating knowledge graphs of the key traffic nodes into global knowledge graphs of the key traffic nodes based on associated relationships of the key traffic nodes within traffic maps corresponding to the large traffic models; andupdating the global knowledge graphs based on status output from traffic prediction results of the large traffic models, so as to generate knowledge graph sets correspondingly; and based on knowledge graph retrieval requests from user terminals, retrieving matched global knowledge graphs from the knowledge graph sets and transmitting the same to the user terminals for visual display.

2. The knowledge graph optimization method for large traffic models according to claim 1, wherein the step of monitoring and filtering traffic data processing tasks of large traffic models to obtain traffic status data of key traffic nodes, and clustering the traffic status data to obtain traffic feature data of the key traffic nodes comprise:based on model network layers where traffic data processing tasks of the large traffic models are located, monitoring entire processes of the traffic data processing tasks to obtain traffic status change data corresponding to the key traffic nodes within target traffic areas when the traffic data processing tasks are executed; and identifying and removing abnormal data from the traffic status change data to obtain traffic status data of the key traffic nodes; andidentifying data types of the traffic status data to obtain data types corresponding to all sub data under the traffic status data; and based on the data types, clustering all sub-data under the traffic status data to obtain traffic feature data of the key traffic nodes, wherein the traffic feature data comprises road network connection feature data, traffic flow feature data and traffic facility feature data of the key traffic nodes.

3. The knowledge graph optimization method for large traffic models according to claim 1, wherein the steps of generating graph layers under knowledge graphs corresponding to the key traffic nodes based on the traffic feature data, andconducting associated identification processing for the graph layers under the knowledge graphs; integrating knowledge graphs of the key traffic nodes into global knowledge graphs of the key traffic nodes based on associated relationships of the key traffic nodes within traffic maps corresponding to the large traffic models comprise:performing metadata identification on the road network connection feature data, the traffic flow feature data, and the traffic facility feature data comprised in the traffic feature data, respectively, to obtain road network connection metadata sets, traffic flow metadata sets, and traffic facility metadata sets correspondingly; and based on the road network connection metadata sets, the traffic flow metadata sets, and the traffic facility metadata sets, respectively constructing road network graph layers, traffic flow graph layers, and traffic facility graph layers under the knowledge graphs corresponding to the key traffic nodes, and conducting associated identification processing for the road network graph layers, the traffic flow graph layers, and the traffic facility graph layers of the key traffic nodes; andbased on positional relationships of the key traffic nodes within the traffic maps corresponding to the large traffic models, determining relative positions and relative distances of the key traffic nodes; and based on the relative positions and the relative distances, integrating the knowledge graphs of the key traffic nodes to obtain global knowledge graphs of the key traffic nodes.

4. The knowledge graph optimization method for large traffic models according to claim 1, wherein the steps of updating the global knowledge graphs based on status output from traffic prediction results of the large traffic models, so as to generate knowledge graph sets correspondingly; and based on knowledge graph retrieval requests from user terminals, retrieving matched global knowledge graphs from the knowledge graph sets and transmitting the same to the user terminals for visual display comprise:obtaining traffic prediction results of the large traffic models at preset time intervals during traffic prediction processes, identifying the traffic prediction results, and determining traffic prediction data of the key traffic nodes comprised in the traffic prediction results; and based on the traffic prediction data, updating knowledge graph portions of the key traffic nodes corresponding to the global knowledge graphs, so as to generate knowledge graph sets correspondingly, wherein updated global knowledge graphs comprised in the knowledge graph sets correspond to the traffic prediction results one by one ; andparsing knowledge graph retrieval requests from user terminals and determining key traffic node attribute information to which knowledge graphs expected by the user terminals belong; and based on the key traffic node attributeinformation, retrieving matched global knowledge graphs from the knowledge graph sets, packaging retrieved matched global knowledge graphs and then transmitting the same to the user terminals for visual display.

5. The knowledge graph optimization method for large traffic models according to claim 4, wherein the step of packaging retrieved matched global knowledge graphs and then transmitting the same to the user terminals for visual display comprises:SI: using a following formula (1), based on displayed widths and heights of the matched global knowledge graphs and displayed widths and heights of the user terminals, to determine initial scaling ratios of the matched global knowledge graphs,where, n represents initial scaling ratios of the matched global knowledgeDv                                 Ugraphs, represents displayed length values of the user terminals, v represents displayed width values of the user terminals, representsdisplayed length values of the matched global knowledge graphs, M represents displayed width values of the matched global knowledge graphs,represents taking minimum values on both sides of a comma enclosed in parentheses;S2: using a following formula (2), based on display ratios of the user terminals and total numbers of pixels at maximum displayable resolutions, as well as display ratios of the matched global knowledge graphs and total numbers of pixels at current resolutions, to determine numbers of pixels to be removed from rows and columns of the matched global knowledge graphs:where represents numbers of vertical columns to be removed from rowsA Vand columns of the matched global knowledge graphs; 1 represents numbersof horizontal columns to be removed from rows and columns of the matchedglobal knowledge graphs; $ represents display ratios of the matched global represents display ratios of the user terminals; , • rr,, A? V - --a ,= — —indicates conditions where 5 and kknowledge graphs;hold true; 1 and represent integer values of aspect ratios in display ratiositof the matched global knowledge graphs; and represent integer valuesof aspect ratios in display ratios of the user terminals; representstaking maximum values on both sides of a comma enclosed in brackets; andS3: using a following formula (3), based on numbers of pixels to be removed from rows and columns of the matched global knowledge graphs and displayed widths and heights of the user terminals, to determine final magnification ratios for displaying the matched global knowledge graphs,where, represents final magnification ratios for displaying the matchedC<global knowledge graphs; represents vertical width values of single pixelbblocks of the matched global knowledge graphs; and horizontal width values of single pixel blocks of the matched global knowledge graphs.

6. A knowledge graph optimization system for large traffic models comprising:a traffic status data acquisition module, used for monitoring and filtering traffic data processing tasks of large traffic models, and obtaining traffic status data of key traffic nodes;a traffic feature data generation module, used for clustering the traffic status data and obtaining traffic feature data of the key traffic nodes;a knowledge graph generation module, used for generating graph layers under knowledge graphs corresponding to the key traffic nodes based on the traffic feature data, and performing associated identification processing on the graphlayers under the knowledge graphs;a knowledge graph integration module, used for integrating the knowledge graphs of the key traffic nodes based on associated relationships of the key traffic nodes within traffic maps corresponding to the large traffic models, and obtaining global knowledge graphs of the key traffic nodes;a knowledge graph update module, used for updating the global knowledge graphs based on status output from traffic prediction results of the large traffic models, and generating knowledge graph sets correspondingly; anda knowledge graph retrieval and transmission module, used for retrieving matched global knowledge graphs from the knowledge graph sets based on knowledge graph retrieval requests from user terminals and transmitting the same to the user terminals for visual display.

7. The knowledge graph optimization system for large traffic models according to claim 6, wherein the traffic status data acquisition module, used for monitoring and filtering traffic data processing tasks of large traffic models, and obtaining traffic status data of key traffic nodes, comprises: based on model network layers where traffic data processing tasks of large traffic models are located, monitoring and obtaining traffic status change data corresponding to the key traffic nodes in target traffic areas when the traffic data processing tasks are executed, identifying and eliminating abnormal data from the traffic status change data, and obtaining traffic status data of the key traffic nodes; andthe traffic feature data generation module, used for clustering the traffic status data and obtaining traffic feature data of the key traffic nodes, comprises:identifying data types of the traffic status data to obtain data types corresponding to all sub-data under the traffic status data; based on the data types, clustering all sub-data under the traffic status data to obtain traffic feature data of the key traffic nodes, wherein the traffic feature data comprises road network connection feature data, traffic flow feature data and traffic facility feature data of the key traffic nodes.

8. The knowledge graph optimization system for large traffic models according to claim 6, wherein the knowledge graph generation module, used for generating several graph layers under the knowledge graphs corresponding to the key traffic nodes based on the traffic feature data, and perform associated identification processing on all graph layers under the knowledge graphs, comprises:performing metadata identification on road network connection feature data, traffic flow feature data and traffic facility feature data comprised in the trafficfeature data to obtain corresponding road network connection metadata sets, traffic flow metadata sets and traffic facility metadata sets; based on the road network connection metadata sets, the traffic flow metadata sets and the traffic facility metadata sets, respectively constructing road network graph layers, traffic flow graph layers and the traffic facility graph layers under knowledge graphs corresponding to the key traffic nodes, and the road network map layer, performing associated identification processing of the key traffic nodes on the road network graph layers, the traffic flow graph layers and the traffic facility graph layers; andthe knowledge graph integration module, used for integrating the knowledge graphs of the key traffic nodes based on associated relationships of the key traffic nodes within traffic maps corresponding to the large traffic models, and obtaining global knowledge graphs of the key traffic nodes, comprises:based on positional relationships of the key traffic nodes within the traffic maps corresponding to the large traffic models, determining relative positions and relative distances of the key traffic nodes; then based on the relative positions and the relative distances, integrating knowledge graphs of the key traffic nodes to obtain global knowledge maps of the key traffic nodes.

9. The knowledge graph optimization system for large traffic models according to claim 6, wherein the knowledge graph update module, used for updating the global knowledge graphs based on status output from traffic prediction results of the large traffic models, and generating knowledge graph sets correspondingly, comprises:retrieving all traffic prediction results of large traffic models at each preset time intervals during traffic prediction processes, identifying all traffic prediction results, and determining traffic prediction data of the key traffic nodes comprised in traffic prediction results; based on the traffic prediction data, updating knowledge graph portions of the key traffic nodes corresponding to the global knowledge graphs, and generating knowledge graph sets correspondingly; wherein all updated knowledge graphs comprised in the knowledge graph sets correspond to all traffic prediction results one by one; andthe knowledge graph retrieval and transmission module, used for retrieving matched global knowledge graphs from the knowledge graph sets based on knowledge graph retrieval requests from user terminals and transmitting the same to the user terminals for visual display, comprises:parsing the knowledge graph retrieval requests from user terminals and determining key traffic node attribute information of the knowledge graphs that the user terminals expect to obtain; and based on the key traffic node attributeinformation, retrieving matched global knowledge graphs from the knowledge graph sets, and then packaging matched global knowledge graphs and transmitting the same to the user terminals for visual display.

Images