5G private network health monitoring method and system based on multi-modal data fusion

Abstract

The invention relates to the technical field of network operation and maintenance, and discloses a 5G private network health monitoring method and system based on multi-modal data fusion, and the method comprises the steps: collecting multi-modal data associated with a target private network system; the multi-modal data comprises wireless performance data, equipment inspection data, operation and maintenance work order data and external associated data; based on a space-time correlation model, carrying out space-time correlation on the multi-modal data to obtain fusion data; based on a fault prediction model, generating a fault analysis result according to the fused data; based on a network health assessment model, generating a health assessment result according to the fusion data and the fault analysis result; and generating a visual operation and maintenance monitoring report corresponding to the target private network system according to the fault analysis result and the health assessment result. Visibly, the monitoring comprehensiveness of the 5G private network can be improved by implementing the method and the system, so that the evaluation accuracy of the health condition of the 5G private network system is improved while the 5G private network fault is efficiently and accurately detected.

Description

Technical Field

[0001] This invention relates to the field of network operation and maintenance technology, and in particular to a method and system for health monitoring of 5G private networks based on multimodal data fusion. Background Technology

[0002] As a private network built on fifth-generation mobile communication technology, 5G private networks are widely used in scenarios with stringent requirements for network performance and stability, such as industrial manufacturing and smart parks, due to their ultra-low latency, high reliability, and independent controllability.

[0003] However, existing 5G private network operation and maintenance systems have significant technical limitations. For example, the monitoring data sources of existing operation and maintenance systems are limited, relying solely on wireless signal data automatically reported by base stations, terminals, and other devices. This results in a one-sided monitoring perspective and makes it difficult to comprehensively assess the health of the 5G private network system. Furthermore, existing operation and maintenance systems can only access and process data after receiving a clear alarm signal. When multiple devices alarm simultaneously, it is difficult to quickly and accurately determine the cause of the fault, resulting in the inability to handle faults in a timely and accurate manner.

[0004] Therefore, it is particularly important to propose a technical solution that can improve the comprehensiveness of 5G private network monitoring, thereby facilitating efficient and accurate detection of 5G private network faults and improving the accuracy of health assessment of the 5G private network system. Summary of the Invention

[0005] This invention provides a 5G private network health monitoring method and system based on multimodal data fusion, which can improve the comprehensiveness of 5G private network monitoring, thereby facilitating efficient and accurate detection of 5G private network faults while improving the accuracy of health status assessment of the 5G private network system.

[0006] To address the aforementioned technical problems, the first aspect of this invention discloses a 5G private network health monitoring method based on multimodal data fusion, the method comprising: Collect multimodal data associated with a target private network system; wherein the target private network system is a 5G private network system and includes multiple devices; the multimodal data includes wireless performance data, equipment inspection data, maintenance work order data, and external related data; Based on the spatiotemporal correlation model, spatiotemporal correlation operation is performed on the multimodal data to obtain fused data; Based on the pre-trained fault prediction model, fault analysis results corresponding to the target private network system are generated according to the fused data. Based on the network health assessment model, a health assessment result corresponding to the target private network system is generated according to the fused data and the fault analysis results. Based on the fault analysis results and the health assessment results, a visual operation and maintenance monitoring report corresponding to the target private network system is generated.

[0007] A second aspect of this invention discloses a 5G private network health monitoring system based on multimodal data fusion, the system comprising: The data acquisition module is used to collect multimodal data associated with the target private network system; wherein, the target private network system is a 5G private network system and the target private network system includes multiple devices; the multimodal data includes wireless performance data, equipment inspection data, maintenance work order data and external related data; The spatiotemporal correlation module is used to perform spatiotemporal correlation operations on the multimodal data based on the spatiotemporal correlation model to obtain fused data; The fault prediction module is used to generate fault analysis results corresponding to the target private network system based on the pre-trained fault prediction model and the fused data. The health assessment module is used to generate a health assessment result for the target private network system based on the network health assessment model, the fused data, and the fault analysis results. The report generation module is used to generate a visual operation and maintenance monitoring report corresponding to the target private network system based on the fault analysis results and the health assessment results.

[0008] As an optional implementation, in a second aspect of the present invention, the spatiotemporal correlation module performs spatiotemporal correlation operations on the multimodal data based on a spatiotemporal correlation model to obtain fused data in the following specific ways: Based on the timestamp alignment algorithm corresponding to the spatiotemporal correlation model, the time points of each data item in the multimodal data are dynamically matched to obtain time-aligned data; Based on the GIS technology corresponding to the spatiotemporal correlation model, and according to the geographical location information of each device in the target private network system, a regional correlation operation is performed on the multimodal data to obtain spatial correlation data; Based on the spatiotemporal correlation between the time-aligned data and the spatially correlated data, fused data is generated.

[0009] As an optional implementation, in the second aspect of the present invention, the specific method by which the fault prediction module generates the fault analysis results corresponding to the target private network system based on the pre-trained fault prediction model and the fused data includes: A feature extraction operation is performed on the fused data to obtain feature extraction results; wherein, the feature extraction operation is used to extract time-series features related to system health from the fused data; The feature extraction results are input into a pre-trained fault prediction model to output fault prediction results for the target private network system; wherein, the fault prediction results include faulty equipment, fault information corresponding to the faulty equipment, and the predicted occurrence time corresponding to the fault information; wherein, the fault information includes fault type and the predicted risk score corresponding to the fault type; Fault information with a predicted risk score higher than or equal to a preset risk score threshold in the fault prediction results is identified as predicted faults, and spatiotemporal correlation data associated with the predicted faults is filtered out from the fused data; wherein, the predicted faults include potential faults and / or obvious faults; Based on the knowledge graph corresponding to the fault prediction model, the predicted fault and the spatiotemporal correlation data are analyzed to obtain the fault cause analysis results corresponding to the predicted fault. The fault analysis results corresponding to the target private network system include at least the fault prediction results; or, the fault analysis results corresponding to the target private network system include at least the fault prediction results and also the fault cause analysis results.

[0010] As an optional implementation, in the second aspect of the present invention, the specific method by which the health assessment module generates the health assessment result corresponding to the target private network system based on the network health assessment model, according to the fused data and the fault analysis result, includes: Based on the fused data, the basic assessment information corresponding to the target private network system is determined; the basic assessment information includes multiple basic assessment sub-information, each of which is one of signal quality information, equipment status information, and service quality information; Based on the network health assessment model, a dynamic weight system corresponding to the target private network system is constructed; wherein, the dynamic weight system consists of multiple target weights, and each target weight is one of the target signal quality weight, target device status weight, and target service quality weight; Based on the dynamic weighting system, the health score corresponding to the target private network system is calculated according to the basic evaluation information. Based on the health score and the fault analysis results, the risk warning level of the target private network system is determined; The health assessment results include a health score and a risk warning level.

[0011] As an optional implementation, in the second aspect of the present invention, the specific method by which the health assessment module constructs the dynamic weight system corresponding to the target private network system based on the network health assessment model includes: Obtain the basic weight system corresponding to the network health assessment model; wherein the basic weight system consists of multiple basic weights, and each basic weight is one of the following: basic signal quality weight, basic equipment status weight, and basic service quality weight; Based on the first weight dynamic adjustment condition corresponding to the network health assessment model, the priority assessment information corresponding to the target private network system is determined from the basic assessment information; Based on the threshold set for the weight of the priority evaluation information in the first weight dynamic adjustment condition, the basic weight corresponding to the priority evaluation information is adjusted to obtain the first optimized weight corresponding to the priority evaluation information. Based on the first optimized weight, the remaining basic weights in the basic weight system are adjusted to obtain the second optimized weight; Based on the first optimized weight and all the second optimized weights, a dynamic weight system corresponding to the target private network system is constructed.

[0012] As an optional implementation, in a second aspect of the present invention, the specific method by which the health assessment module determines the priority assessment information corresponding to the target private network system from the basic assessment information based on the first weight dynamic adjustment conditions corresponding to the network health assessment model includes: Determine the application scenario corresponding to the target private network system; Based on the application scenario, the business analysis results of the target private network system under the application scenario are determined; the business analysis results include core business and the business risk types corresponding to the core business; Based on the business analysis results, assess the correlation between each basic assessment sub-information in the basic assessment information and the business analysis results; The basic evaluation sub-information with a correlation degree higher than or equal to a preset correlation threshold is determined as the first priority evaluation information corresponding to the target private network system; And / or, Obtain the dynamic evaluation driving conditions corresponding to the target private network system; Based on the dynamic evaluation driving conditions, the second priority evaluation information corresponding to the target private network system is determined; Based on the first priority assessment information and / or the second priority assessment information, the priority assessment information corresponding to the target private network system is determined.

[0013] As an optional implementation, in the second aspect of the invention, each of the basic evaluation sub-information corresponds to at least two initial dynamic sub-weights; The health assessment module, based on the dynamic weighting system and the basic assessment information, calculates the health score of the target private network system using the following specific methods: For each type of basic evaluation sub-information, determine whether the basic evaluation sub-information meets the pre-set second weight dynamic adjustment condition; For each type of basic evaluation sub-information, when it is determined that the basic evaluation sub-information meets the second weight dynamic adjustment condition, the initial dynamic sub-weight corresponding to the basic evaluation sub-information is adjusted according to the second weight dynamic adjustment condition to obtain the dynamically adjusted sub-weight, which is used as the target sub-weight. For each type of basic evaluation sub-information, when it is determined that the basic evaluation sub-information does not meet the second weight dynamic adjustment condition, the initial dynamic sub-weight corresponding to the basic evaluation sub-information is retained as the target sub-weight; For each of the aforementioned basic assessment sub-information, based on the health score formula corresponding to the basic assessment sub-information, and according to the basic assessment sub-information and the target sub-weight corresponding to the basic assessment sub-information, the health score corresponding to the basic assessment sub-information is calculated. The health score of the target private network system is calculated based on the health score corresponding to each of the basic assessment sub-information and the target weight corresponding to each of the basic assessment sub-information.

[0014] A third aspect of this invention discloses another 5G private network health monitoring system based on multimodal data fusion, the system comprising: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute some or all of the steps of the 5G private network health monitoring method based on multimodal data fusion disclosed in the first aspect of the present invention.

[0015] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute some or all of the steps of the 5G private network health monitoring method based on multimodal data fusion disclosed in the first aspect of the present invention.

[0016] Compared with the prior art, the present invention has the following beneficial effects: In this invention, multimodal data associated with a target private network system is collected. The target private network system is a 5G private network system and includes multiple devices. The multimodal data includes wireless performance data, device inspection data, maintenance work order data, and external correlation data. Based on a spatiotemporal correlation model, spatiotemporal correlation operations are performed on the multimodal data to obtain fused data. Based on a pre-trained fault prediction model, fault analysis results corresponding to the target private network system are generated according to the fused data. Based on a network health assessment model, health assessment results corresponding to the target private network system are generated according to the fused data and fault analysis results. Based on the fault analysis results and health assessment results, a visualized operation and maintenance monitoring report corresponding to the target private network system is generated. As can be seen, implementing this invention enables the collection of multimodal data associated with the target private network system. Based on a spatiotemporal correlation model, the multimodal data is spatiotemporally correlated to obtain fused data. Then, based on the fused data, fault prediction is performed on the target private network system to obtain fault analysis results. Next, based on a network health assessment model, health assessment results are generated according to the fused data and fault analysis results. Finally, based on the fault analysis results and health assessment results, corresponding visualized operation and maintenance monitoring reports are output. This allows for fault prediction and health assessment of 5G private networks based on multimodal data fusion, achieving efficient operation and maintenance monitoring of 5G private networks. This improves the comprehensiveness of 5G private network monitoring, facilitating efficient and accurate fault detection while enhancing the accuracy of health assessments. Furthermore, it reduces the cost of manual inspections of 5G private networks, enables efficient early warning of system faults, and facilitates more timely handling and location of system faults, improving the operational reliability and security of 5G private networks and ultimately enhancing the user experience. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a 5G private network health monitoring method based on multimodal data fusion disclosed in an embodiment of the present invention. Figure 2 This is a flowchart illustrating another 5G private network health monitoring method based on multimodal data fusion disclosed in an embodiment of the present invention. Figure 3 This is a flowchart illustrating a 5G private network health monitoring method based on multimodal data fusion disclosed in an embodiment of the present invention. Figure 4This is a schematic diagram of the structure of a 5G private network health monitoring system based on multimodal data fusion disclosed in an embodiment of the present invention; Figure 5 This is a schematic diagram of another 5G private network health monitoring system based on multimodal data fusion disclosed in an embodiment of the present invention. Detailed Implementation

[0019] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0021] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0022] This invention discloses a 5G private network health monitoring method and system based on multimodal data fusion. It can collect multimodal data associated with a target private network system, perform spatiotemporal correlation on the multimodal data based on a spatiotemporal correlation model to obtain fused data, and then perform fault prediction on the target private network system based on the fused data to obtain fault analysis results. Then, based on a network health assessment model, it generates health assessment results based on the fused data and fault analysis results, and finally outputs corresponding visualized operation and maintenance monitoring reports based on the fault analysis results and health assessment results. This method enables fault prediction and health assessment of 5G private networks based on multimodal data fusion, achieving efficient operation and maintenance monitoring of 5G private networks. This improves the comprehensiveness of 5G private network monitoring, facilitating efficient and accurate fault detection while enhancing the accuracy of health assessment. Furthermore, it reduces the cost of manual inspection of 5G private networks, enables efficient early warning of system faults, and facilitates more timely handling and location of system faults, improving the operational reliability and security of 5G private networks and enhancing the user experience. Detailed explanations follow.

[0023] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating a 5G private network health monitoring method based on multimodal data fusion disclosed in an embodiment of the present invention. Wherein, Figure 1 The described 5G private network health monitoring method based on multimodal data fusion can be applied to a 5G private network health monitoring system based on multimodal data fusion. This system may include one of a smart terminal, a smart device, and a server, wherein the server may be a local server or a cloud server; this embodiment of the invention does not limit the specific implementation. Figure 1 As shown, the 5G private network health monitoring method based on multimodal data fusion may include the following operations: 101. Collect multimodal data associated with the target private network system.

[0024] In this embodiment of the invention, the target private network system is a 5G private network system, and the target private network system may include multiple devices; wherein, each device may be one of a base station, communication infrastructure equipment and service equipment, and this embodiment of the invention does not limit the types of devices.

[0025] In this embodiment of the invention, multimodal data may include wireless performance data, equipment inspection data, maintenance work order data, and externally related data. Optionally, the data types included in the multimodal data may include one or more combinations of voice data, image data, text data, video data, and sensor data, and this embodiment of the invention does not limit this. Further optionally, wireless performance data may include, but is not limited to, RSRP (Reference Signal Received Power), RSRQ (Reference Signal Received Quality), and SINR (Signal to Interference plus Noise) reported by any of the above-mentioned devices. Ratio (signal-to-interference-plus-noise ratio), throughput, latency, packet loss rate, connection establishment success rate, etc., may also include device operation data (e.g., CPU utilization). This embodiment of the invention is not limited to these parameters. Optionally, device inspection data may include, but is not limited to, image data and / or video data about the appearance of any device periodically collected by a camera or inspection robot. This embodiment of the invention is not limited to these parameters. Optionally, maintenance work order data may include, but is not limited to, maintenance log data and / or assurance record data reported by users through at least one of voice, image, text, and video. This embodiment of the invention is not limited to these parameters. Optionally, external related data may include, but is not limited to, external environmental data, power supply status, and production scheduling status of the business system corresponding to the target private network system. This embodiment of the invention is not limited to these parameters.

[0026] In this embodiment of the invention, for example, the collection of multimodal data associated with the target private network system may specifically include: voice fault reporting, that is, collecting fault phenomena described by users through voice (such as "video lag"); image acquisition, that is, periodically taking photos of the appearance of the equipment (such as base station antenna); and equipment monitoring, that is, real-time acquisition of wireless signal quality (such as SINR value) and equipment operating status (such as CPU utilization). This embodiment of the invention does not limit the scope of the invention.

[0027] 102. Based on the spatiotemporal correlation model, perform spatiotemporal correlation operations on multimodal data to obtain fused data.

[0028] 103. Based on the pre-trained fault prediction model, generate fault analysis results for the target private network system according to the fused data.

[0029] In this embodiment of the invention, optionally, the fault prediction model can be obtained by improving upon Long Short-Term Memory (LSTM) or Support Vector Machine (SVM), and this embodiment of the invention does not limit it.

[0030] 104. Based on the network health assessment model, generate the health assessment results corresponding to the target private network system according to the fused data and fault analysis results.

[0031] In this embodiment of the invention, the health assessment result may optionally include a health score.

[0032] 105. Based on the fault analysis results and health assessment results, generate a visual operation and maintenance monitoring report corresponding to the target private network system.

[0033] In this embodiment of the invention, optionally, the visualized operation and maintenance monitoring report may include a three-dimensional health map corresponding to the health assessment results; however, this embodiment of the invention does not impose any limitations on this.

[0034] As can be seen, the method described in the embodiments of the present invention can collect multimodal data associated with the target private network system, perform spatiotemporal correlation on the multimodal data based on a spatiotemporal correlation model to obtain fused data, and then perform fault prediction on the target private network system based on the fused data to obtain fault analysis results. Then, based on a network health assessment model, a health assessment result is generated based on the fused data and the fault analysis result. Finally, based on the fault analysis result and the health assessment result, a corresponding visualized operation and maintenance monitoring report is output. This method can achieve fault prediction and health assessment for 5G private networks based on multimodal data fusion, realize efficient operation and maintenance monitoring of 5G private networks, and improve the comprehensiveness of 5G private network monitoring. This is beneficial for achieving efficient and accurate detection of 5G private network faults, while improving the accuracy of health assessment of the 5G private network system. Furthermore, it is beneficial for reducing the cost of manual inspection of 5G private networks, achieving efficient early warning of system faults, and facilitating more timely handling and location of system faults, thereby improving the operational reliability and security of 5G private networks and enhancing the user experience of 5G private networks.

[0035] In an optional embodiment, performing a spatiotemporal correlation operation on multimodal data based on a spatiotemporal correlation model to obtain fused data may include the following operations: Based on the timestamp alignment algorithm corresponding to the spatiotemporal correlation model, the time points of various data in the multimodal data are dynamically matched to obtain time-aligned data; Based on the GIS (Geographic Information System) technology corresponding to the spatiotemporal correlation model, regional correlation operations are performed on the multimodal data according to the geographical location information of each device in the target private network system to obtain spatial correlation data; Based on the spatiotemporal correlation between time-aligned data and spatially correlated data, fused data is generated.

[0036] Optionally, the geographic location information may include one or more combinations of latitude and longitude, three-dimensional coordinates, and region / grid code, which are not limited in this embodiment of the invention.

[0037] Spatial association data is used to represent the association of multimodal data with corresponding geographic location information, so as to facilitate subsequent spatiotemporal correlation of the data.

[0038] Optionally, the above-mentioned generation of fused data based on the spatiotemporal correlation between time-aligned data and spatially correlated data can be specifically as follows: Define a configurable time window (e.g., ±500 milliseconds) and a configurable geospatial extent (e.g., a geographic circle with a radius of 50 meters or a logical polygon region). Clustering operations are performed on time-aligned data and spatially correlated data to obtain multiple spatiotemporal events; the data included in each spatiotemporal event are all within the same time window and the same spatial range; The data included in each spatiotemporal event are fused to integrate the heterogeneous data included in that spatiotemporal event into structured data, thus obtaining fused data.

[0039] As can be seen, this optional embodiment can align data from different sources by timestamp and geographic location, and generate fused data through spatiotemporal correlation, which can improve the accuracy of processing multimodal data, thereby improving the efficiency of subsequent data analysis based on multimodal data for fault prediction and health assessment.

[0040] In an optional embodiment, based on a pre-trained fault prediction model and fused data, generating fault analysis results corresponding to the target private network system may include the following operations: A feature extraction operation is performed on the fused data to obtain the feature extraction results; the feature extraction operation is used to extract time-series features related to system health from the fused data; The feature extraction results are input into a pre-trained fault prediction model to output the fault prediction results corresponding to the target private network system. The fault prediction results include the faulty device, the fault information corresponding to the faulty device, and the predicted occurrence time corresponding to the fault information. The fault information includes the fault type and the predicted risk score corresponding to the fault type. Fault information with a predicted risk score higher than or equal to a preset risk score threshold in the fault prediction results is identified as predicted faults, and spatiotemporal correlation data associated with the predicted faults are filtered out from the fused data; wherein, predicted faults include potential faults and / or obvious faults. Based on the knowledge graph corresponding to the fault prediction model, we analyze the predicted faults and spatiotemporal correlation data to obtain the fault cause analysis results corresponding to the predicted faults. The fault analysis results corresponding to the target private network system include at least the fault prediction results; or, the fault analysis results corresponding to the target private network system include at least the fault prediction results and also the fault cause analysis results.

[0041] The predicted risk score can be used to represent the confidence level of the fault type, but this embodiment of the invention does not limit it.

[0042] Among them, the predicted risk score for obvious faults is higher than that for potential faults.

[0043] Optionally, the fault cause analysis results may include at least one fault cause and the confidence level of each fault cause, which is not limited in this embodiment of the invention; further optionally, the fault cause may include equipment information corresponding to at least one faulty device, wherein the equipment information may include, but is not limited to, one or more of equipment identifier, equipment type, equipment location information, and equipment operating status information, which is not limited in this embodiment of the invention.

[0044] Among them, the knowledge graph is used to record the total causal relationships between various types of multimodal data corresponding to the target private network system. In this way, the graph algorithm can be used to find the path with the closest connection and the most complete evidence chain between the predicted fault and the knowledge graph, and then find the cause of the fault corresponding to the predicted fault.

[0045] For example, in the collected multimodal data, when the voice report mentions "AGV vehicle response delay", and the image acquisition shows that there is a crack in the base station shell, and the equipment monitoring data shows that the wireless signal fluctuation in the area is abnormal, the cause of the fault can be comprehensively judged as "poor heat dissipation of the base station leading to performance degradation". This embodiment of the invention does not limit the scope.

[0046] As can be seen, this optional embodiment can first extract the features of the fused data and input the feature extraction results into the trained fault prediction model to output the fault prediction results. Then, by analyzing the correlation between multimodal data, the cause of the fault can be determined, which can improve the efficiency and accuracy of fault prediction and more efficiently and accurately determine the cause of the fault. This is conducive to more efficient handling and repair of the fault in the future, and thus helps to improve the operational reliability of the target private network system.

[0047] Example 2 Please see Figure 2 , Figure 2 This is a flowchart illustrating a 5G private network health monitoring method based on multimodal data fusion disclosed in an embodiment of the present invention. Wherein, Figure 2 The described 5G private network health monitoring method based on multimodal data fusion can be applied to a 5G private network health monitoring system based on multimodal data fusion. This system may include one of a smart terminal, a smart device, and a server, wherein the server may be a local server or a cloud server; this embodiment of the invention does not limit the specific implementation. Figure 2 As shown, the 5G private network health monitoring method based on multimodal data fusion may include the following operations: 201. Collect multimodal data associated with the target private network system.

[0048] In this embodiment of the invention, the target private network system is a 5G private network system, and the target private network system includes multiple devices; the multimodal data includes wireless performance data, device inspection data, maintenance work order data, and external related data.

[0049] 202. Based on the spatiotemporal correlation model, perform spatiotemporal correlation operations on multimodal data to obtain fused data.

[0050] 203. Based on the pre-trained fault prediction model, generate fault analysis results for the target private network system according to the fused data.

[0051] 204. Based on the fusion data, determine the basic assessment information corresponding to the target private network system.

[0052] In this embodiment of the invention, the basic assessment information includes multiple basic assessment sub-information, each of which is one of signal quality information, device status information, and service quality information; wherein, signal quality information can be used to reflect the wireless transmission reliability of the target private network system, device status information can be used to characterize the hardware operating health status of the target private network system, and service quality information can be used to assess the service continuity assurance capability of the target private network system.

[0053] 205. Based on the network health assessment model, construct a dynamic weight system corresponding to the target private network system.

[0054] In this embodiment of the invention, the dynamic weighting system consists of multiple target weights, and each target weight is one of the target signal quality weight, target device status weight, and target service quality weight; wherein, the sum of all target weights can be equal to 1.

[0055] 206. Based on the dynamic weighting system and the basic assessment information, calculate the health score corresponding to the target private network system.

[0056] 207. Based on the health score and fault analysis results, determine the risk warning level of the target private network system.

[0057] In this embodiment of the invention, optionally, determining the risk warning level of the target private network system based on the health score and fault analysis results may include the following operations: Based on the health score, the first risk warning level of the target private network system is determined; the first risk warning level can be one of the green warning level, yellow warning level, and red warning level. Based on the fault analysis results, the second risk warning level of the target private network system is determined; the second risk warning level can be one of the green warning level, yellow warning level, and red warning level. When the first risk warning level is detected to be consistent with the second risk warning level, the first risk warning level will be determined as the risk warning level of the target private network system. When a discrepancy is detected between the first risk warning level and the second risk warning level, the warning level that indicates the higher degree of risk severity between the first and second risk warning levels shall be determined as the risk warning level of the target private network system; or, based on historical health assessment results, historical fault analysis results, and historical fault operation and maintenance records, the first confidence level corresponding to the first risk warning level and the second confidence level corresponding to the second risk warning level shall be analyzed. When the first confidence level is greater than or equal to the second confidence level, the first risk warning level shall be determined as the risk warning level of the target private network system; when the first confidence level is less than the second confidence level, the second risk warning level shall be determined as the risk warning level of the target private network system.

[0058] The green alert level indicates that the target private network system is in good health and only requires continuous monitoring; the yellow alert level indicates that the target private network system has potential risks; and the red alert level indicates that the target private network system has obvious risks and the risks / faults must be dealt with immediately.

[0059] Further optionally, determining the first risk warning level of the target private network system based on the health score may include the following operations: When the health score is detected to be greater than or equal to the first preset score threshold, the first risk warning level of the target private network system is determined to be the green warning level. When the health score is detected to be less than the first preset score threshold and the health score is greater than or equal to the second preset score threshold, the first risk warning level of the target private network system is determined to be the yellow warning level; wherein, the first preset score threshold is greater than the second preset score threshold. When the health score is detected to be less than the second preset score threshold, the first risk warning level of the target private network system is determined to be the red warning level.

[0060] For example, the first preset scoring threshold can be 80 points, and the second preset scoring threshold can be 70 points. This embodiment of the invention does not limit the score.

[0061] Further, optionally, based on the fault analysis results, determining the second risk warning level of the target private network system may include the following operations: When the fault analysis results indicate that there are no faults in the target private network system, the second risk warning level of the target private network system is determined to be the green warning level. When the fault analysis results are used to indicate that the target private network system has a predicted fault, the second risk warning level of the target private network system is determined to be the yellow warning level. When the fault analysis results indicate that there is a significant fault in the target private network system, the second risk warning level of the target private network system is determined to be the red warning level.

[0062] 208. Based on the fault analysis results and health assessment results, generate a visual operation and maintenance monitoring report corresponding to the target private network system.

[0063] In this embodiment of the invention, the health assessment result includes a health score and a risk warning level; the risk warning level includes a first risk warning level or a second risk warning level.

[0064] In this embodiment of the invention, optionally, the visualized operation and maintenance monitoring report may include a fault root cause tracing tree corresponding to the fault analysis results and a three-dimensional health map corresponding to the health assessment results. This embodiment of the invention does not impose any limitations. Further optionally, the fault root cause tracing tree can be used to graphically display the fault propagation path (e.g., base station overheating → CPU frequency reduction → signal quality degradation → AGV control delay), and can also support clicking to view related evidence (e.g., the base station infrared thermal imaging map when the early warning is triggered). This embodiment of the invention does not impose any limitations. Further optionally, the three-dimensional health map may include an X-axis, a Y-axis, and a Z-axis, wherein the X-axis can be a timeline (supporting 24-hour / 7-day / 30-day view switching), the Y-axis can be the health score in the health assessment results, and the Z-axis can be a multi-dimensional indicator radar chart (multi-dimensional indicators may include signal dimension, equipment dimension, and service dimension). Based on this, the three-dimensional health map can visualize the changing trend of the health scores of the multi-dimensional indicators to reflect the changes in the health status of the private network system corresponding to the multi-dimensional indicators. This embodiment of the invention does not impose any limitations.

[0065] For further detailed descriptions of steps 201-203 and 208 in this embodiment of the invention, please refer to the detailed descriptions of steps 101-103 and 105 in Embodiment 1. These descriptions will not be repeated in this embodiment of the invention.

[0066] As can be seen, the method described in the embodiments of the present invention can collect multimodal data associated with the target private network system, perform spatiotemporal correlation on the multimodal data based on a spatiotemporal correlation model to obtain fused data, and then perform fault prediction on the target private network system based on the fused data to obtain fault analysis results. Then, based on a network health assessment model, a health assessment result is generated based on the fused data and the fault analysis result. Finally, based on the fault analysis result and the health assessment result, a corresponding visualized operation and maintenance monitoring report is output. This method can achieve fault prediction and health assessment for 5G private networks based on multimodal data fusion, realize efficient operation and maintenance monitoring of 5G private networks, and improve the comprehensiveness of 5G private network monitoring. This is beneficial for achieving efficient and accurate detection of 5G private network faults, while improving the accuracy of health assessment of the 5G private network system. Furthermore, it is beneficial for reducing the cost of manual inspection of 5G private networks, achieving efficient early warning of system faults, and facilitating more timely handling and location of system faults, thereby improving the operational reliability and security of 5G private networks and enhancing the user experience of 5G private networks. Furthermore, based on the fused data, it can determine basic assessment information such as signal quality information, equipment status information, and service quality information. Based on a network health assessment model, it constructs a dynamic weighting system corresponding to the target private network system. Then, based on this dynamic weighting system and combined with the basic assessment information, it calculates the health score of the target private network system. Finally, by combining the health score with fault analysis results, it determines the risk warning level of the target private network system. The dynamic weighting system enables intelligent adjustment of the scoring benchmark for 5G private networks, thereby improving the flexibility and accuracy of health score calculation, and thus enhancing the flexibility in determining risk warning levels. This allows for more sensitive assessment of whether fault repair or inspection of the 5G private network is necessary, ultimately improving the accuracy and reliability of 5G private network operation and maintenance.

[0067] In an optional embodiment, constructing a dynamic weight system corresponding to the target private network system based on the network health assessment model may include the following operations: Obtain the basic weight system corresponding to the network health assessment model; wherein, the basic weight system consists of multiple basic weights, and each basic weight is one of the following: basic signal quality weight, basic equipment status weight, and basic service quality weight; Based on the first weight dynamic adjustment condition corresponding to the network health assessment model, the priority assessment information corresponding to the target private network system is determined from the basic assessment information. Based on the threshold set for the weight of priority evaluation information in the first weight dynamic adjustment condition, the basic weight corresponding to the priority evaluation information is adjusted to obtain the first optimized weight corresponding to the priority evaluation information. Based on the first optimized weight, the remaining basic weights in the basic weight system are adjusted to obtain the second optimized weight; Based on the first optimization weight and all the second optimization weights, a dynamic weight system corresponding to the target private network system is constructed.

[0068] Specifically, adjusting the base weights corresponding to the priority evaluation information involves increasing the base weights corresponding to the priority evaluation information so that the adjusted first optimization weight is higher than the base weights.

[0069] The weight setting threshold for priority evaluation information is used to restrict the first optimized weight obtained after adjustment to be less than or equal to the weight setting threshold. For example, when the priority evaluation information is business quality information, the weight setting threshold can be 50%; when the priority evaluation information is device status information, the weight setting threshold can be 45%; the weight setting threshold can also be set to other values, which are not limited in this embodiment of the invention.

[0070] Each target weight can be one of the first optimization weight and all the second optimization weights, and this embodiment of the invention does not limit it.

[0071] As can be seen, this optional embodiment can first obtain the basic weight system corresponding to the network health assessment model, and then determine the priority assessment information corresponding to the target private network system from the basic assessment information according to the first weight dynamic adjustment condition, so as to adjust the basic weight corresponding to the priority assessment information to obtain the corresponding first optimized weight, thereby adaptively adjusting the remaining basic weights to obtain the second optimized weight, so as to construct the dynamic weight system corresponding to the target private network system. This can realize the dynamic adjustment of the score assessment weight of network health assessment, thereby improving the flexibility and accuracy of the construction of the dynamic weight system, and further improving the accuracy of the health score calculation for 5G private networks.

[0072] In this optional embodiment, the process of determining the priority assessment information corresponding to the target private network system from the basic assessment information based on the first weight dynamic adjustment condition corresponding to the network health assessment model may include the following operations: Determine the application scenarios corresponding to the target private network system; Based on the application scenario, determine the business analysis results of the target private network system under the application scenario; the business analysis results include core business and the business risk types corresponding to the core business; Based on the business analysis results, assess the correlation between each basic assessment sub-information in the basic assessment information and the business analysis results; Basic evaluation sub-information with a correlation degree higher than or equal to a preset correlation degree threshold is identified as the first priority evaluation information corresponding to the target private network system; And / or, Obtain the dynamic evaluation driving conditions corresponding to the target private network system; Based on dynamic evaluation driving conditions, determine the second priority evaluation information corresponding to the target private network system; Based on the first priority assessment information and / or the second priority assessment information, determine the priority assessment information corresponding to the target private network system.

[0073] For example, the application scenarios corresponding to the target private network system may include, but are not limited to, one of the following: industrial manufacturing, smart parks, intelligent transportation, and medical IoT. This embodiment of the invention does not impose any limitations on these scenarios. Furthermore, the core services of the target private network system in these application scenarios may respectively include AGV (Automated Guided Vehicle) scheduling in industrial manufacturing, IoT device collaboration in smart parks, autonomous driving communication in intelligent transportation, and remote device monitoring in medical IoT. Moreover, the corresponding business risk types for the core services in different application scenarios may be: in the industrial manufacturing scenario, the business risk type may be: network interruption or latency jitter directly leading to production stoppages, equipment collisions, and product scrapping; in the smart park scenario, the business risk type may be: localized failures affecting user experience and management efficiency; in the intelligent transportation scenario, the business risk type may be: communication interruptions or bit errors may cause traffic accidents, endangering lives; in the medical IoT scenario, the business risk type may be: data loss or delays may delay diagnosis or treatment, directly affecting patient safety. This embodiment of the invention does not impose any limitations on these scenarios.

[0074] The higher the correlation between any basic assessment sub-information and the business analysis results, the greater the impact of that basic assessment sub-information on the core business in that application scenario, and the easier it is to influence the core business in that application scenario to generate / avoid business risks.

[0075] For example, the dynamic evaluation driving condition can be a service priority driving condition or an environment state driving condition, and this embodiment of the invention is not limited thereto; further, based on the dynamic evaluation driving condition, determining the second priority evaluation information corresponding to the target private network system can specifically be as follows: When the dynamic evaluation driving condition is the service priority driving condition, if a high-security-level service (such as industrial control instructions) is detected in the target private network system, the second priority evaluation information corresponding to the target private network system is determined to be the service quality information. When the dynamic evaluation driving condition is the environmental state driving condition, if the target private network system is detected to be in an extreme environment (e.g., excessive temperature / humidity) based on external correlation data, the second priority evaluation information corresponding to the target private network system is determined to be the device status information. This embodiment of the invention does not limit this.

[0076] Optionally, determining the priority evaluation information corresponding to the target private network system based on the first priority evaluation information and / or the second priority evaluation information may include the following operations: The first priority evaluation information or the second priority evaluation information can be directly determined as the priority evaluation information corresponding to the target private network system; or, the intersection / union between the first priority evaluation information and the second priority evaluation information can be determined as the priority evaluation information of the target private network system. This embodiment of the invention does not limit the specific choices.

[0077] As can be seen, this optional embodiment can also determine the core business and corresponding business risk types of the target private network system in the application scenario corresponding to the target private network system. Thus, the basic assessment sub-information with a high correlation with the above business analysis results in the basic assessment information can be determined as the first priority assessment information. Alternatively, the second priority assessment information corresponding to the target private network system can be determined based on dynamic assessment driving conditions. Then, priority assessment information can be determined based on the first priority assessment information and / or the second priority assessment information. This can provide different screening methods for priority assessment information, thereby improving the flexibility and accuracy of determining priority assessment information. This, in turn, helps to improve the adjustment flexibility and accuracy of the weight system corresponding to the health score, and thus helps to improve the accuracy of the health score.

[0078] In this optional embodiment, each basic evaluation sub-information may correspond to at least two initial dynamic sub-weights; wherein the sum of all initial dynamic sub-weights corresponding to each basic evaluation sub-information may be equal to 1.

[0079] Optionally, when the basic evaluation sub-information is signal quality information, the initial dynamic sub-weights corresponding to the signal quality information may include signal performance index weights and fluctuation stability weights; when the basic evaluation sub-information is device status information, the initial dynamic sub-weights corresponding to the device status information may include hardware health weights and appearance integrity weights; when the basic evaluation sub-information is service quality information, the initial dynamic sub-weights corresponding to the service quality information may include service compliance weights and service perception weights. This embodiment of the invention does not impose any limitations.

[0080] In this optional embodiment, optionally, calculating the health score corresponding to the target private network system based on the dynamic weighting system and the basic assessment information may include the following operations: For each type of basic evaluation sub-information, determine whether the basic evaluation sub-information meets the pre-set second weight dynamic adjustment condition; For each type of basic evaluation sub-information, when it is determined that the basic evaluation sub-information meets the second weight dynamic adjustment condition, the initial dynamic sub-weight corresponding to the basic evaluation sub-information is adjusted according to the second weight dynamic adjustment condition to obtain the dynamically adjusted sub-weight, which is used as the target sub-weight. For each type of basic evaluation sub-information, when it is determined that the basic evaluation sub-information does not meet the second weight dynamic adjustment condition, the initial dynamic sub-weight corresponding to the basic evaluation sub-information is retained as the target sub-weight; For each basic assessment sub-information, based on the health score formula corresponding to the basic assessment sub-information, and according to the basic assessment sub-information and the target sub-weights corresponding to the basic assessment sub-information, the health score corresponding to the basic assessment sub-information is calculated. The health score of the target private network system is calculated based on the health score corresponding to each basic assessment sub-information and the target weight corresponding to each basic assessment sub-information.

[0081] Optionally, for each basic evaluation sub-information, determining whether the basic evaluation sub-information meets the pre-set second weight dynamic adjustment condition may include the following operations: When the basic evaluation sub-information is signal quality information, it is determined whether the network load of the target private network system exceeds the preset load value (e.g., 70%). When it is determined that the network load of the target private network system exceeds the preset load value, it is determined that the signal quality information meets the second weight dynamic adjustment condition. When the basic evaluation sub-information is device status information, it is determined whether the device type currently required for analysis of the target private network system has changed; when it is determined that the device type currently required for analysis of the target private network system has changed, it is determined that the device status information meets the second weight dynamic adjustment condition. When the basic assessment sub-information is service quality information, it is determined whether the priority coefficient corresponding to the current service of the target private network system is higher than the preset priority; when it is determined that the priority coefficient corresponding to the current service of the target private network system is higher than the preset priority, it is determined that the service quality information meets the second weight dynamic adjustment condition.

[0082] Further optionally, for each basic evaluation sub-information, the initial dynamic sub-weight corresponding to that basic evaluation sub-information is adjusted according to the second weight dynamic adjustment condition to obtain the dynamically adjusted sub-weight, which serves as the target sub-weight. This can include the following operations: When the basic evaluation sub-information is signal quality information, the fluctuation stability weight corresponding to the signal quality information is increased, and the signal performance index weight corresponding to the signal quality information is adjusted accordingly, which is used as the target sub-weight corresponding to the signal quality information. When the basic evaluation sub-information is device status information, the hardware health weight (e.g., the hardware health weight for a base station can be 0.8, and the hardware health weight for a switch can be 0.6) and appearance integrity weight are dynamically adjusted according to the type of device to be analyzed, and used as the target sub-weights for the device status information. When the basic assessment sub-information is business quality information, the business compliance weight corresponding to the business quality information is increased, and the business perception weight corresponding to the business quality information is adjusted accordingly, which serves as the target sub-weight corresponding to the business quality information.

[0083] Further optionally, for each basic assessment sub-information, based on the health score formula corresponding to that basic assessment sub-information, and according to that basic assessment sub-information and its corresponding target sub-weight, the health score corresponding to that basic assessment sub-information is calculated, which may include the following operations: When the basic evaluation sub-information is signal quality information, the compliance rate of the signal performance indicators (such as SINR / RSRP / PDCP layer latency) and the network fluctuation stability value of the target private network system are determined based on the signal quality information. Based on the health scoring formula corresponding to the signal quality information, the signal quality score is calculated according to the aforementioned compliance rate, network fluctuation stability value, and target sub-weights, and serves as the health score corresponding to the signal quality information. The health scoring formula corresponding to this signal quality information can be as follows: S = (Signal performance index compliance rate × weight α) + (Network fluctuation stability value × weight β); Where S is the signal quality score, weight α is the weight of the above signal performance index, and weight β is the weight of the above fluctuation stability. When the basic assessment sub-information is equipment status information, based on the equipment status information, the hardware health of the target private network system is determined by fusing data from one or more sensors, including temperature, voltage, and fan speed. The severity of physical defects detected by visual inspection is also determined, serving as the equipment appearance integrity of the target private network system. Based on the health scoring formula corresponding to the equipment status information, and according to the aforementioned hardware health, equipment appearance integrity, and target sub-weights, an equipment status score is calculated, which serves as the health score corresponding to the equipment status information. The health scoring formula corresponding to this equipment status information can be as follows: D = (Hardware health × weight γ) + (Device appearance integrity × weight δ); Where D is the device status score, weight γ is the aforementioned hardware health weight, and weight δ is the aforementioned appearance integrity weight. When the basic assessment sub-information is service quality information, the basic indicator scores (such as latency / packet loss rate / throughput) of the target private network system and the service perception score based on user evaluation are determined according to the service quality information. Based on the health score formula corresponding to the service quality information, the service quality score is calculated according to the above basic indicator scores, service perception scores, and target sub-weights, and is used as the health score corresponding to the service quality information. The health score formula corresponding to the service quality information can be as follows: Q = (Basic indicator score × weight ε) + (Business perception score × weight ζ); Where Q is the business quality score, weight ε is the business compliance weight, and weight ζ is the business perception weight.

[0084] As can be seen, this optional embodiment can also, for each basic assessment sub-information, when it is determined that the basic assessment sub-information meets the second weight dynamic adjustment condition, adjust the initial dynamic sub-weight corresponding to the basic assessment sub-information according to the second weight dynamic adjustment condition to obtain the target sub-weight; otherwise, directly use the initial dynamic sub-weight as the target sub-weight. Then, based on the health score formula, according to the basic assessment sub-information and the determined target sub-weight, calculate the health score corresponding to the basic assessment sub-information. Then, combining the health score corresponding to each basic assessment sub-information and the target weight corresponding to each basic assessment sub-information, calculate the health score corresponding to the target private network system. This can improve the flexibility of weight adjustment for each basic assessment sub-information, thereby improving the scoring flexibility and accuracy of each basic assessment sub-information, which is conducive to further improving the calculation accuracy of the health score.

[0085] In this embodiment of the invention, the exemplary flowchart of the 5G private network health monitoring method based on multimodal data fusion can be found in [reference needed]. Figure 3 , Figure 3 This is a flowchart illustrating a 5G private network health monitoring method based on multimodal data fusion disclosed in an embodiment of the present invention; wherein the method may include the following operations: First, multimodal data associated with the target private network system is collected through the voice fault reporting system, visual acquisition unit, and equipment performance detection unit in the multi-source data acquisition layer. Then, through the multimodal fusion engine, the multimodal data undergoes spatiotemporal correlation operations such as dynamic time warping and graph neural network association to obtain fused data. Next, through the intelligent analysis layer, the fused data is analyzed to achieve fault prediction and root cause analysis, resulting in fault analysis results. Then, through the health assessment and early warning layer, the target private network system is scored based on dynamic weights, and early warning classification is performed based on the health score. Finally, the fault analysis results, health scores, and classification results are combined and visualized. This embodiment of the invention is not limited.

[0086] Example 3 Please see Figure 4 , Figure 4 This is a schematic diagram of the structure of a 5G private network health monitoring system based on multimodal data fusion, as disclosed in an embodiment of the present invention. Wherein, Figure 4 The described 5G private network health monitoring system based on multimodal data fusion may include one of a smart terminal, a smart device, and a server, wherein the server may be a local server or a cloud server, and this embodiment of the invention is not limited thereto. Figure 4 As shown, the 5G private network health monitoring system based on multimodal data fusion may include: The data acquisition module 301 is used to collect multimodal data associated with the target private network system; wherein, the target private network system is a 5G private network system and includes multiple devices; the multimodal data includes wireless performance data, equipment inspection data, maintenance work order data and external related data; The spatiotemporal correlation module 302 is used to perform spatiotemporal correlation operations on multimodal data based on the spatiotemporal correlation model to obtain fused data; The fault prediction module 303 is used to generate fault analysis results corresponding to the target private network system based on the pre-trained fault prediction model and the fused data. The health assessment module 304 is used to generate health assessment results for the target private network system based on the network health assessment model, fused data, and fault analysis results. The report generation module 305 is used to generate a visual operation and maintenance monitoring report for the target private network system based on the fault analysis results and health assessment results.

[0087] As can be seen, the system described in the embodiments of this invention can collect multimodal data associated with the target private network system, perform spatiotemporal correlation on the multimodal data based on a spatiotemporal correlation model to obtain fused data, and then perform fault prediction on the target private network system based on the fused data to obtain fault analysis results. Then, based on a network health assessment model, it generates health assessment results based on the fused data and fault analysis results, and outputs corresponding visualized operation and maintenance monitoring reports based on the fault analysis results and health assessment results. It can realize fault prediction and health assessment for 5G private networks based on multimodal data fusion, achieve efficient operation and maintenance monitoring of 5G private networks, thereby improving the comprehensiveness of 5G private network monitoring. This is beneficial for achieving efficient and accurate detection of 5G private network faults, while improving the accuracy of health assessment of the 5G private network system. In turn, it is beneficial for reducing the cost of manual inspection of 5G private networks, achieving efficient early warning of system faults, and facilitating more timely handling and location of system faults, thereby improving the operational reliability and security of 5G private networks and enhancing the user experience of 5G private networks.

[0088] In an optional embodiment, the spatiotemporal correlation module 302 performs spatiotemporal correlation operations on multimodal data based on a spatiotemporal correlation model to obtain the specific method of fused data, which may include: Based on the timestamp alignment algorithm corresponding to the spatiotemporal correlation model, the time points of various data in the multimodal data are dynamically matched to obtain time-aligned data; Based on the GIS technology corresponding to the spatiotemporal correlation model, regional correlation operations are performed on the multimodal data according to the geographical location information of each device in the target private network system to obtain spatial correlation data; Based on the spatiotemporal correlation between time-aligned data and spatially correlated data, fused data is generated.

[0089] As can be seen, the system described in this optional embodiment can align data from different sources by timestamp and geographic location, and generate fused data through spatiotemporal correlation, which can improve the accuracy of processing multimodal data, thereby improving the efficiency of subsequent data analysis based on multimodal data for fault prediction and health assessment.

[0090] In an optional embodiment, the specific method by which the fault prediction module 303 generates fault analysis results corresponding to the target private network system based on a pre-trained fault prediction model and fused data may include: A feature extraction operation is performed on the fused data to obtain the feature extraction results; the feature extraction operation is used to extract time-series features related to system health from the fused data; The feature extraction results are input into a pre-trained fault prediction model to output the fault prediction results corresponding to the target private network system. The fault prediction results include the faulty device, the fault information corresponding to the faulty device, and the predicted occurrence time corresponding to the fault information. The fault information includes the fault type and the predicted risk score corresponding to the fault type. Fault information with a predicted risk score higher than or equal to a preset risk score threshold in the fault prediction results is identified as predicted faults, and spatiotemporal correlation data associated with the predicted faults are filtered out from the fused data; wherein, predicted faults include potential faults and / or obvious faults. Based on the knowledge graph corresponding to the fault prediction model, we analyze the predicted faults and spatiotemporal correlation data to obtain the fault cause analysis results corresponding to the predicted faults. The fault analysis results corresponding to the target private network system include at least the fault prediction results; or, the fault analysis results corresponding to the target private network system include at least the fault prediction results and also the fault cause analysis results.

[0091] As can be seen, the system described in this optional embodiment can first extract features from the fused data and input the feature extraction results into the trained fault prediction model to output fault prediction results. Then, by analyzing the correlation between multimodal data, the cause of the fault can be determined, which can improve the efficiency and accuracy of fault prediction and more efficiently and accurately determine the cause of the fault. This is conducive to more efficient handling and repair of faults in the future, and thus helps to improve the operational reliability of the target private network system.

[0092] In an optional embodiment, the specific method by which the health assessment module 304 generates the health assessment result corresponding to the target private network system based on the network health assessment model, fused data, and fault analysis results may include: Based on the fused data, the basic assessment information corresponding to the target private network system is determined. The basic assessment information includes a variety of basic assessment sub-information, each of which is one of signal quality information, equipment status information, and service quality information. Based on the network health assessment model, a dynamic weight system corresponding to the target private network system is constructed. The dynamic weight system consists of multiple target weights, and each target weight is one of the target signal quality weight, target device status weight, and target service quality weight. Based on a dynamic weighting system and basic assessment information, the health score of the target private network system is calculated. Based on the health score and fault analysis results, the risk warning level of the target private network system is determined; The health assessment results include a health score and a risk warning level.

[0093] As can be seen, the system described in this optional embodiment can determine basic assessment information such as signal quality information, device status information, and service quality information based on fused data. Based on a network health assessment model, it constructs a dynamic weighting system corresponding to the target private network system. Then, based on the dynamic weighting system and combined with the basic assessment information, it calculates the health score of the target private network system. Finally, by combining the health score and fault analysis results, it determines the risk warning level of the target private network system. The dynamic weighting system enables intelligent adjustment of the scoring benchmark for 5G private networks, thereby improving the flexibility and accuracy of health score calculation, and thus increasing the flexibility in determining the risk warning level. This allows for more sensitive assessment of whether fault repair or inspection of the 5G private network is necessary, ultimately improving the accuracy and reliability of 5G private network operation and maintenance.

[0094] In this optional embodiment, the specific method by which the health assessment module 304 constructs the dynamic weight system corresponding to the target private network system based on the network health assessment model may include: Obtain the basic weight system corresponding to the network health assessment model; wherein, the basic weight system consists of multiple basic weights, and each basic weight is one of the following: basic signal quality weight, basic equipment status weight, and basic service quality weight; Based on the first weight dynamic adjustment condition corresponding to the network health assessment model, the priority assessment information corresponding to the target private network system is determined from the basic assessment information. Based on the threshold set for the weight of priority evaluation information in the first weight dynamic adjustment condition, the basic weight corresponding to the priority evaluation information is adjusted to obtain the first optimized weight corresponding to the priority evaluation information. Based on the first optimized weight, the remaining basic weights in the basic weight system are adjusted to obtain the second optimized weight; Based on the first optimization weight and all the second optimization weights, a dynamic weight system corresponding to the target private network system is constructed.

[0095] As can be seen, the system described in this optional embodiment can first obtain the basic weight system corresponding to the network health assessment model, and then determine the priority assessment information corresponding to the target private network system from the basic assessment information according to the first weight dynamic adjustment condition, so as to adjust the basic weight corresponding to the priority assessment information to obtain the corresponding first optimized weight, thereby adaptively adjusting the remaining basic weights to obtain the second optimized weight, so as to construct the dynamic weight system corresponding to the target private network system. This can realize the dynamic adjustment of the score assessment weight of network health assessment, thereby improving the flexibility and accuracy of the construction of the dynamic weight system, and further improving the accuracy of the health score calculation for 5G private networks.

[0096] In this optional embodiment, optionally, the specific method by which the health assessment module 304 determines the priority assessment information corresponding to the target private network system from the basic assessment information based on the first weight dynamic adjustment conditions corresponding to the network health assessment model may include: Determine the application scenarios corresponding to the target private network system; Based on the application scenario, determine the business analysis results of the target private network system under the application scenario; the business analysis results include core business and the business risk types corresponding to the core business; Based on the business analysis results, assess the correlation between each basic assessment sub-information in the basic assessment information and the business analysis results; Basic evaluation sub-information with a correlation degree higher than or equal to a preset correlation degree threshold is identified as the first priority evaluation information corresponding to the target private network system; And / or, Obtain the dynamic evaluation driving conditions corresponding to the target private network system; Based on dynamic evaluation driving conditions, determine the second priority evaluation information corresponding to the target private network system; Based on the first priority assessment information and / or the second priority assessment information, determine the priority assessment information corresponding to the target private network system.

[0097] As can be seen, the system described in this optional embodiment can also determine the core business and corresponding business risk types of the target private network system in the application scenario, based on the application scenario corresponding to the target private network system. This allows the basic assessment sub-information with a high correlation to the above business analysis results to be identified as the first priority assessment information. Alternatively, it can determine the second priority assessment information corresponding to the target private network system based on dynamic assessment driving conditions. Then, based on the first and / or second priority assessment information, priority assessment information is determined. This provides different screening methods for priority assessment information, thereby improving the flexibility and accuracy of determining priority assessment information. This, in turn, helps improve the adjustment flexibility and accuracy of the weighting system corresponding to the health score, ultimately improving the accuracy of the health score.

[0098] In this optional embodiment, each basic evaluation sub-information may correspond to at least two initial dynamic sub-weights; Among them, the health assessment module 304, based on a dynamic weighting system, calculates the health score of the target private network system according to the basic assessment information, and may include the following specific methods: For each type of basic evaluation sub-information, determine whether the basic evaluation sub-information meets the pre-set second weight dynamic adjustment condition; For each type of basic evaluation sub-information, when it is determined that the basic evaluation sub-information meets the second weight dynamic adjustment condition, the initial dynamic sub-weight corresponding to the basic evaluation sub-information is adjusted according to the second weight dynamic adjustment condition to obtain the dynamically adjusted sub-weight, which is used as the target sub-weight. For each type of basic evaluation sub-information, when it is determined that the basic evaluation sub-information does not meet the second weight dynamic adjustment condition, the initial dynamic sub-weight corresponding to the basic evaluation sub-information is retained as the target sub-weight; For each basic assessment sub-information, based on the health score formula corresponding to the basic assessment sub-information, and according to the basic assessment sub-information and the target sub-weights corresponding to the basic assessment sub-information, the health score corresponding to the basic assessment sub-information is calculated. The health score of the target private network system is calculated based on the health score corresponding to each basic assessment sub-information and the target weight corresponding to each basic assessment sub-information.

[0099] As can be seen, the system described in this optional embodiment can also, for each basic assessment sub-information, adjust the initial dynamic sub-weight corresponding to the basic assessment sub-information according to the second weight dynamic adjustment condition when it is determined that the basic assessment sub-information meets the second weight dynamic adjustment condition, to obtain the target sub-weight; otherwise, the initial dynamic sub-weight is directly used as the target sub-weight. Then, based on the health score formula, the health score corresponding to the basic assessment sub-information is calculated according to the basic assessment sub-information and the determined target sub-weight. Then, combining the health score corresponding to each basic assessment sub-information and the target weight corresponding to each basic assessment sub-information, the health score corresponding to the target private network system is calculated. This can improve the flexibility of weight adjustment for each basic assessment sub-information, thereby improving the scoring flexibility and accuracy of each basic assessment sub-information, which is conducive to further improving the calculation accuracy of the health score.

[0100] Example 4 Please see Figure 5 , Figure 5 This is a schematic diagram of another 5G private network health monitoring system based on multimodal data fusion disclosed in an embodiment of the present invention. Figure 5 As shown, the 5G private network health monitoring system based on multimodal data fusion may include: Memory 401 storing executable program code; Processor 402 coupled to memory 401; The processor 402 calls the executable program code stored in the memory 401 to execute some or all of the steps in the 5G private network health monitoring method based on multimodal data fusion described in Embodiment 1 or Embodiment 2 of the present invention.

[0101] Example 5 This invention discloses a computer storage medium storing computer instructions. When these computer instructions are invoked, they are used to execute some or all of the steps in the 5G private network health monitoring method based on multimodal data fusion described in Embodiment 1 or Embodiment 2 of this invention.

[0102] Example 6 This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform some or all of the steps in the 5G private network health monitoring method based on multimodal data fusion described in Embodiment 1 or Embodiment 2.

[0103] The system embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0104] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0105] Finally, it should be noted that the 5G private network health monitoring method and system based on multimodal data fusion disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for health monitoring of 5G private networks based on multimodal data fusion, characterized in that, The method includes: Collect multimodal data associated with a target private network system; wherein the target private network system is a 5G private network system and includes multiple devices; the multimodal data includes wireless performance data, equipment inspection data, maintenance work order data, and external related data; Based on the spatiotemporal correlation model, spatiotemporal correlation operation is performed on the multimodal data to obtain fused data; Based on the pre-trained fault prediction model, fault analysis results corresponding to the target private network system are generated according to the fused data. Based on the network health assessment model, a health assessment result corresponding to the target private network system is generated according to the fused data and the fault analysis results. Based on the fault analysis results and the health assessment results, a visual operation and maintenance monitoring report corresponding to the target private network system is generated.

2. The 5G private network health monitoring method based on multimodal data fusion according to claim 1, characterized in that, The spatiotemporal correlation model is used to perform spatiotemporal correlation operations on the multimodal data to obtain fused data, including: Based on the timestamp alignment algorithm corresponding to the spatiotemporal correlation model, the time points of each data item in the multimodal data are dynamically matched to obtain time-aligned data; Based on the GIS technology corresponding to the spatiotemporal correlation model, regional correlation operations are performed on the multimodal data according to the geographical location information of each device in the target private network system to obtain spatial correlation data; Based on the spatiotemporal correlation between the time-aligned data and the spatially correlated data, fused data is generated.

3. The 5G private network health monitoring method based on multimodal data fusion according to claim 1, characterized in that, The fault analysis results generated based on the pre-trained fault prediction model and the fused data for the target private network system include: A feature extraction operation is performed on the fused data to obtain feature extraction results; wherein, the feature extraction operation is used to extract time-series features related to system health from the fused data; The feature extraction results are input into a pre-trained fault prediction model to output fault prediction results for the target private network system; wherein, the fault prediction results include faulty equipment, fault information corresponding to the faulty equipment, and the predicted occurrence time corresponding to the fault information; wherein, the fault information includes fault type and the predicted risk score corresponding to the fault type; Fault information with a predicted risk score higher than or equal to a preset risk score threshold in the fault prediction results is identified as predicted faults, and spatiotemporal correlation data associated with the predicted faults is filtered out from the fused data; wherein, the predicted faults include potential faults and / or obvious faults; Based on the knowledge graph corresponding to the fault prediction model, the predicted fault and the spatiotemporal correlation data are analyzed to obtain the fault cause analysis results corresponding to the predicted fault. The fault analysis results corresponding to the target private network system include at least the fault prediction results; or, the fault analysis results corresponding to the target private network system include at least the fault prediction results and also the fault cause analysis results.

4. The 5G private network health monitoring method based on multimodal data fusion according to any one of claims 1-3, characterized in that, The network health assessment model generates a health assessment result for the target private network system based on the fused data and the fault analysis results, including: Based on the fused data, the basic assessment information corresponding to the target private network system is determined; the basic assessment information includes multiple basic assessment sub-information, each of which is one of signal quality information, equipment status information, and service quality information; Based on the network health assessment model, a dynamic weight system corresponding to the target private network system is constructed; wherein, the dynamic weight system consists of multiple target weights, and each target weight is one of the target signal quality weight, target device status weight, and target service quality weight; Based on the dynamic weighting system, the health score corresponding to the target private network system is calculated according to the basic evaluation information. Based on the health score and the fault analysis results, the risk warning level of the target private network system is determined; The health assessment results include a health score and a risk warning level.

5. The 5G private network health monitoring method based on multimodal data fusion according to claim 4, characterized in that, The dynamic weighting system for the target private network system, constructed based on the network health assessment model, includes: Obtain the basic weight system corresponding to the network health assessment model; wherein, the basic weight system consists of multiple basic weights, and each of the basic weights is one of the following: basic signal quality weight, basic equipment status weight, and basic service quality weight; Based on the first weight dynamic adjustment condition corresponding to the network health assessment model, the priority assessment information corresponding to the target private network system is determined from the basic assessment information; Based on the weight set threshold for the priority evaluation information in the first weight dynamic adjustment condition, the basic weight corresponding to the priority evaluation information is adjusted to obtain the first optimized weight corresponding to the priority evaluation information. Based on the first optimized weight, the remaining basic weights in the basic weight system are adjusted to obtain the second optimized weight; Based on the first optimized weight and all the second optimized weights, a dynamic weight system corresponding to the target private network system is constructed.

6. The 5G private network health monitoring method based on multimodal data fusion according to claim 5, characterized in that, The step of determining the priority assessment information corresponding to the target private network system from the basic assessment information based on the first weight dynamic adjustment condition corresponding to the network health assessment model includes: Determine the application scenario corresponding to the target private network system; Based on the application scenario, the business analysis results of the target private network system under the application scenario are determined; the business analysis results include core business and the business risk types corresponding to the core business; Based on the business analysis results, assess the correlation between each of the basic assessment sub-information in the basic assessment information and the business analysis results; The basic evaluation sub-information with a correlation degree higher than or equal to a preset correlation threshold is determined as the first priority evaluation information corresponding to the target private network system; And / or, Obtain the dynamic evaluation driving conditions corresponding to the target private network system; Based on the dynamic evaluation driving conditions, the second priority evaluation information corresponding to the target private network system is determined; Based on the first priority assessment information and / or the second priority assessment information, the priority assessment information corresponding to the target private network system is determined.

7. The 5G private network health monitoring method based on multimodal data fusion according to claim 5 or 6, characterized in that, Each of the aforementioned basic evaluation sub-information corresponds to at least two initial dynamic sub-weights; The step of calculating the health score of the target private network system based on the dynamic weighting system and the basic evaluation information includes: For each type of basic evaluation sub-information, determine whether the basic evaluation sub-information meets the pre-set second weight dynamic adjustment condition; For each type of basic evaluation sub-information, when it is determined that the basic evaluation sub-information meets the second weight dynamic adjustment condition, the initial dynamic sub-weight corresponding to the basic evaluation sub-information is adjusted according to the second weight dynamic adjustment condition to obtain the dynamically adjusted sub-weight, which is used as the target sub-weight. For each type of basic evaluation sub-information, when it is determined that the basic evaluation sub-information does not meet the second weight dynamic adjustment condition, the initial dynamic sub-weight corresponding to the basic evaluation sub-information is retained as the target sub-weight; For each of the aforementioned basic assessment sub-information, based on the health score formula corresponding to the basic assessment sub-information, and according to the basic assessment sub-information and the target sub-weight corresponding to the basic assessment sub-information, the health score corresponding to the basic assessment sub-information is calculated. The health score of the target private network system is calculated based on the health score corresponding to each of the basic assessment sub-information and the target weight corresponding to each of the basic assessment sub-information.

8. A 5G private network health monitoring system based on multimodal data fusion, characterized in that, The system includes: A data acquisition module is used to collect multimodal data associated with a target private network system; wherein the target private network system is a 5G private network system and includes multiple devices; the multimodal data includes wireless performance data, equipment inspection data, maintenance work order data, and external related data; The spatiotemporal correlation module is used to perform spatiotemporal correlation operations on the multimodal data based on the spatiotemporal correlation model to obtain fused data; The fault prediction module is used to generate fault analysis results corresponding to the target private network system based on the pre-trained fault prediction model and the fused data. The health assessment module is used to generate a health assessment result for the target private network system based on the network health assessment model, the fused data, and the fault analysis results. The report generation module is used to generate a visual operation and maintenance monitoring report corresponding to the target private network system based on the fault analysis results and the health assessment results.

9. A 5G private network health monitoring system based on multimodal data fusion, characterized in that, The system includes: Memory containing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the 5G private network health monitoring method based on multimodal data fusion as described in any one of claims 1-7.

10. A computer storage medium, characterized in that, The computer storage medium stores computer instructions, which, when invoked, are used to execute the 5G private network health monitoring method based on multimodal data fusion as described in any one of claims 1-7.

Images