Fault route rapid positioning system based on AI

Abstract

The invention, which relates to the technical field of the wireless communication network, discloses an AI-based fault route rapid positioning system comprising: a whole network flow topology data synchronous acquisition module used for obtaining synchronized original data from network elements in the wireless communication network; the disturbance characteristic time sequence extraction module is used for generating a characteristic time sequence based on the synchronized original data; the flow topology coupling effect quantification module is used for combining the characteristic time sequence with the network topology index data to generate a real-time fault fingerprint sequence; and the fault source reverse reasoning module is used for acquiring a preset fault fingerprint database, calculating the fingerprint difference degree and judging the prototype fingerprint with the minimum difference degree so as to generate a fault positioning report. According to the method, the fault source can be globally and accurately positioned, the soft fault can be sensitively detected, the positioning precision and robustness are improved by fusing multi-dimensional information, and adaptive parameter optimization and learning expansion can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of wireless communication networks, and in particular to an AI-based fault routing rapid location system. Background Art

[0002] With the deployment of next-generation wireless communication networks, network scale, complexity, and dynamism have increased dramatically. Large-scale networks feature a vast number of base stations and core network elements, with complex connections. A single network element failure can often cause local or global performance disturbances across the entire network through traffic rerouting and user handoffs. These disturbances exhibit complex propagation characteristics across time and space, posing significant challenges to fault location.

[0003] Existing wireless network fault location methods rely heavily on the timeliness and accuracy of alarms, relying on correlation analysis of network element alarms. These methods are weak at locating "soft" faults that don't generate obvious alarms, such as external interference, improper parameter configuration, and hidden capacity bottlenecks. Furthermore, static correlation rules struggle to adapt to dynamic changes in network topology and service patterns, leading to false positives and missed detections. Anomaly detection methods based on single-base station KPIs operate from a siloed perspective, only identifying the fault location but not the cause. For example, a degradation in service quality at a certain base station may be caused by user handoff congestion caused by a fault at an adjacent base station. Traditional methods struggle to infer the root cause from local data. Existing technologies lack a global mechanism for tracing faults by leveraging the spatiotemporal patterns formed by fault disturbances propagating through service flows and network topology, making it difficult to address the technical challenges of large-scale network operations and maintenance. Summary of the Invention

[0004] The purpose of the present invention is to provide an AI-based fault routing rapid location system to solve the problems existing in the background technology.

[0005] To solve the above technical problems, the present invention provides an AI-based fault routing rapid location system, comprising: a full-network flow topology data synchronization acquisition module, for synchronously collecting service flow indicator data and network topology indicator data from network elements in a wireless communication network to obtain synchronized raw data; The disturbance feature temporal extraction module is used to calculate the time change rate and short-term volatility of service flow indicator data based on the synchronized raw data to generate a characteristic time series that represents the network status disturbance; a flow topology coupling effect quantification module, configured to combine the characteristic time series with the network topology indicator data to quantify the mutual influence strength between pairs of nodes in the network, so as to generate a real-time fault fingerprint sequence characterizing the mutual influence strength; The fault root cause reverse reasoning module is used to obtain a preset fault fingerprint library, which contains prototype fingerprints corresponding to multiple known fault types; calculate the fingerprint difference between the real-time fault fingerprint sequence and each prototype fingerprint in the fault fingerprint library; and determine the prototype fingerprint with the smallest fingerprint difference. The fault type and root cause node corresponding to the prototype fingerprint with the smallest fingerprint difference are used as the positioning result to generate a fault location report.

[0006] Preferably, the disturbance feature temporal extraction module is specifically used to: Performing first-order difference calculation on the throughput time series in the business flow indicator data to obtain a throughput change rate series; For the time series of the number of active users in the business flow indicator data, calculate the standard deviation of the sliding window to obtain the volatility series of the number of active users; The throughput change rate series and the active user number volatility series are combined to generate the characteristic time series.

[0007] Preferably, the flow topology coupling effect quantification module is specifically used to: Calculating the normalized flow disturbance correlation of the node pair; Calculating the dynamic topology impact factor of the node pair; Performing a weighted summation on the normalized flow disturbance correlation and the dynamic topology impact factor to obtain a flow topology coupling index; A coupling effect matrix is constructed based on the flow topology coupling index to generate the real-time fault fingerprint sequence.

[0008] Preferably, the calculation of the normalized flow disturbance correlation is specifically as follows: Sliding the characteristic time series of the node pair within a preset time delay range, and calculating the Pearson correlation coefficient between the two series for each delay time point; The maximum absolute value of the correlation coefficient among all the delayed time points is selected as the normalized flow disturbance correlation.

[0009] Preferably, the dynamic topology impact factor is calculated as follows: Obtaining the number of successful handovers between the node pair in the current time window and dividing it by the historical average number of handovers to obtain a normalized handover behavior factor; Obtaining a reference signal received power from the source node reported by the user equipment of the target node, and subtracting a preset signal quality reference threshold to obtain a signal quality difference factor; multiplying the normalized handover behavior factor by a preset handover impact weight, and multiplying the signal quality difference factor by a preset signal quality impact weight; The two products are nonlinearly fused through a hyperbolic tangent function to generate the dynamic topology influence factor.

[0010] Preferably, the switching impact weight and the signal quality impact weight are hyperparameters obtained through machine learning model training and optimization based on historical fault case data.

[0011] Preferably, the fingerprint difference is calculated as follows: Calculating the Frobenius norm between the real-time coupling effect matrix and the coupling effect matrix of the prototype fingerprint at each time point within a preset fingerprint sequence length; The arithmetic mean of the Frobenius norms at all time points is calculated to obtain the fingerprint difference.

[0012] Preferably, the fault root cause reverse reasoning module is further used to: Compare the calculated minimum fingerprint difference with a preset confidence threshold; When the minimum fingerprint difference is lower than the confidence threshold, outputting the positioning result; When the minimum fingerprint difference is not lower than the confidence threshold, a report indicating that the fault type is unknown is output.

[0013] Preferably, the confidence threshold is determined by drawing a receiver operating characteristic curve on the validation dataset and based on preset positioning accuracy and recall rate targets.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. By establishing a synchronous collection module for network-wide flow topology data, the system is no longer limited to a single network element, but synchronously collects service flow indicator data and network topology indicator data across the entire network from a global perspective. On this basis, the flow topology coupling effect quantification module calculates the flow topology coupling index and constructs a coupling effect matrix, generating a real-time fault fingerprint sequence that can depict the dynamic impact relationship between nodes in the entire network, achieving globalization and root cause positioning, and breaking through the limitations of local perspectives.

[0015] 2. Highly sensitive to "soft" faults and independent of hardware alarms, the disturbance feature temporal extraction module shifts the focus from the absolute value of performance indicators to their subtle dynamic changes by calculating the time rate of change and short-term volatility. This module detects and locates subtle dynamic disturbances in service and network performance, improving the ability to locate soft faults without warnings.

[0016] 3. By integrating the quantitative coupling effect of business flows and network topology heterogeneous information, the constructed fault fingerprint sequence has a higher dimension and richer information content than a single information source. Its recognition far exceeds that of a single type of indicator, thereby significantly improving the accuracy and robustness of positioning.

[0017] 4. Key system parameters are adaptively optimized through machine learning, and the fault fingerprint library is continuously learned and expanded, with the ability to adapt to network evolution and new fault modes. This enables the entire positioning system to adapt to network evolution and the emergence of new fault modes, and has the ability to continuously evolve, ensuring its long-term effectiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention, and those skilled in the art can derive other drawings based on these drawings without inventive effort. Figure 1 This is a logic block diagram of the AI-based fault routing rapid location system of the present invention; Figure 2 This is the calculation flow chart of the dynamic topology impact factor; Figure 3 The calculation flow chart of fingerprint difference. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0020] Example 1: See also Figures 1 to 3 , the present invention provides an AI-based fault routing rapid location system, comprising: a full-network flow topology data synchronization acquisition module, for synchronously collecting service flow indicator data and network topology indicator data from network elements in a wireless communication network to obtain synchronized original data; A disturbance feature time series extraction module is used to calculate the time change rate and short-term volatility of the service flow indicator data based on the synchronized original data to generate a characteristic time series representing the network state disturbance; a flow topology coupling effect quantification module, configured to combine the characteristic time series with the network topology indicator data to quantify the mutual influence strength between pairs of nodes in the network, so as to generate a real-time fault fingerprint sequence characterizing the mutual influence strength; A fault root cause reverse reasoning module is used to obtain a preset fault fingerprint library containing prototype fingerprints corresponding to multiple known fault types; calculate the fingerprint difference between the real-time fault fingerprint sequence and each prototype fingerprint in the fault fingerprint library; determine the prototype fingerprint with the smallest fingerprint difference, and use the fault type and root cause node corresponding to the prototype fingerprint with the smallest fingerprint difference as the location result to generate a fault location report; The embodiments of the present invention provide a systematic fault location framework. First, through the full-network flow topology data synchronization acquisition module, network dynamics are captured from a global perspective, which avoids the limitations of traditional methods that rely solely on isolated alarms or single network element performance indicators. Then, the disturbance feature time-series extraction module converts the original performance data into features that can better reflect changes in network status, laying the foundation for in-depth analysis. Its core lies in the flow topology coupling effect quantification module, which innovatively constructs fault fingerprints and reveals the complex propagation patterns of faults in the two dimensions of business flow and network topology. Finally, the fault root cause reverse reasoning module achieves accurate tracing from phenomenon to root cause by comparing with preset typical fault patterns. The value of this framework lies in its ability to penetrate the chain reaction caused by the fault, directly identify the root cause of the problem, and achieve root cause location of network faults.

[0021] Example 2: The disturbance feature temporal extraction module is specifically used for: Performing first-order difference calculation on the throughput time series in the business flow indicator data to obtain a throughput change rate series; For the time series of the number of active users in the business flow indicator data, calculate the standard deviation of the sliding window to obtain the volatility series of the number of active users; and combining the throughput change rate series and the active user number volatility series to generate the characteristic time series; The flow topology coupling effect quantification module is specifically used for: Calculating the normalized flow disturbance correlation of the node pair; Calculating the dynamic topology impact factor of the node pair; Performing a weighted summation on the normalized flow disturbance correlation and the dynamic topology impact factor to obtain a flow topology coupling index; and constructing a coupling effect matrix based on the flow topology coupling index to generate the real-time fault fingerprint sequence; In this embodiment, the disturbance feature temporal extraction module operates by converting raw performance metrics into feature metrics that are more sensitive to changes. It performs first-order difference calculations on throughput, converting absolute value metrics into rate-of-change metrics. This transformation enables the system to more sensitively capture transient performance fluctuations. Furthermore, by calculating the sliding window standard deviation of the number of active users, it effectively quantifies the short-term volatility of user behavior. The subtlety of this processing approach lies in filtering out the stable portion of the performance metric and amplifying the dynamic information that characterizes network disturbances. The flow topology coupling effect quantification module integrates this disturbance information with network topology changes to construct a comprehensive impact metric. It calculates the flow topology coupling index and uses it to construct a coupling effect matrix, generating a real-time fault fingerprint sequence that describes the correlation relationships across the entire network. This is significant because it no longer views the performance of each node in isolation, but rather quantifies the strength of the mutual influence between nodes when a fault occurs from the perspective of the entire network. This provides high-dimensional, highly recognizable input for subsequent fault tracing, significantly improving location accuracy. The core of this module is to calculate the flow topology coupling index, which is defined as follows: ; represents the flow topology coupling index, a dimensionless scalar value that indicates the coupling strength between the disturbance state of node i and the disturbance state of node j at time point t. The larger the value, the closer the correlation between the two and the more likely the disturbance has propagated between them. represents the flow weight and topology weight, two dimensionless positive constants, and These are adjustable hyperparameters of the system, used to balance the importance of business flow factors and network topology factors in the calculation of the coupling index. They are initially set based on expert experience and can be optimized through subsequent machine learning model training. represents the normalized flow disturbance correlation, a dimensionless scalar value that quantifies the synchronization or delay correlation of service flow disturbances between nodes i and j; represents the dynamic topology impact factor, a dimensionless scalar value that quantifies the tightness or drastic change of the topological connection between nodes i and j at time point t; The subscripts i and j represent two different nodes in the network, such as the base station in the following text; The subscript t represents the time point of the current calculation; The technical motivation for this formula is that neither simple service flow correlation nor topology correlation is sufficient to fully describe fault propagation. A fault first occurs at the topological level, such as signal degradation, or at the service level, such as link interruption. Then, through the inherent mechanisms of the network, it propagates across both levels. For example, a failure at node i may cause a degradation in the service flow quality of users under its coverage, a flow disturbance. These users will attempt to switch to a neighboring node j, a topological event, which in turn triggers service flow congestion at node j, a new flow disturbance. This formula creatively combines the propagation of flow disturbances with the influence of topological structure through a weighted fusion, aiming to capture this complex causal transmission effect and thus construct a more recognizable fault fingerprint. This module periodically calculates the , thus generating a coupling effect matrix at each time snapshot t , where the element in the i-th row and j-th column of the matrix is ; When a network failure occurs, the coupling index between the affected node pairs will change. A unique, high-dimensional dynamic pattern is formed in the time series; this dynamic matrix sequence is the "flow-topology" disturbance propagation fingerprint defined in this invention. "Flow-topology" is the abbreviation of business flow and network topology. The dynamic coupling relationship between the two is the core basis for the reverse reasoning of the fault root cause in this invention; the output of this module - the fault fingerprint sequence , which will serve as the direct input to the final decision module; Hyperparameters in the system, including And in the subsequent formula , can be optimized through supervised learning methods; specifically, using a labeled data set containing known fault cases and their root causes, the accuracy of the final fault location is used as the optimization target, and algorithms such as gradient descent are used to reversely adjust these parameters to find the optimal combination; specifically, since the accuracy is not differentiable, a differentiable proxy loss function is usually used in model training, such as the cross-entropy loss function; the system regards the fault location problem as a multi-classification task (classifying the current fault fingerprint into a prototype fingerprint in the library), and drives the gradient descent algorithm to minimize the cross entropy between the fault type probability distribution predicted by the model and the actual fault label. Other configuration parameters such as Etc., are set based on prior knowledge of network engineering to ensure the rationality of their physical meaning.

[0022] Example 3: The calculation of the normalized flow disturbance correlation is specifically as follows: Sliding the characteristic time series of the node pair within a preset time delay range, and calculating the Pearson correlation coefficient between the two series for each delay time point; and selecting the maximum absolute value of the correlation coefficient among all the delayed time points as the normalized flow disturbance correlation; The calculation of the dynamic topology impact factor is specifically as follows: Obtaining the number of successful handovers between the node pair in the current time window and dividing it by the historical average number of handovers to obtain a normalized handover behavior factor; Obtaining a reference signal received power from the source node reported by the user equipment of the target node, and subtracting a preset signal quality reference threshold to obtain a signal quality difference factor; multiplying the normalized handover behavior factor by a preset handover impact weight, and multiplying the signal quality difference factor by a preset signal quality impact weight; and performing nonlinear fusion on the two products through a hyperbolic tangent function to generate the dynamic topology impact factor; This example further refines the calculation method of the two core components of the flow topology coupling index; Normalized flow disturbance correlation The calculation of is intended to capture the temporal causal relationship of business flow disturbances; its calculation formula is: ; Indicates that node i is in the time window Time series of business flow disturbance characteristics within; Indicates that node j has passed time delay The characteristic time series after sliding, its time window is , both sequences are derived from the output of the disturbance feature temporal extraction module; The time window size is a system configuration parameter that defines the length of the time series considered when calculating the correlation. It is set based on network service characteristics and fault response time, for example, 5 minutes. Indicates time delay, one in The variable to search within the range to find the time difference corresponding to the strongest correlation between two disturbance series; The maximum search delay is a system configuration parameter that represents the maximum possible time for disturbance propagation in the network. It is estimated based on the physical size of the network and the protocol processing delay, for example, 2 seconds. represents the Pearson correlation coefficient function, which is a standard statistical function well known in the art and is used to measure the degree of linear correlation between two variables. Its output value is between [-1, 1]; The significance of this formula is to find the strongest correlation between the traffic disturbances of two nodes within a reasonable delay range, capturing both positive correlations (such as congestion transmission) and negative correlations (such as traffic flow fluctuations). Dynamic topology impact factor The calculation of quantifies the association strength and changes at the network topology level; its calculation formula is: ; : dynamic topology impact factor, a dimensionless scalar; : Hyperbolic tangent function; : are the handover impact weight and signal quality impact weight, respectively, which are dimensionless adjustable hyperparameters; : Normalized switching behavior factor, obtained by comparing the current number of switching times with the historical average; : Normalized signal quality factor, dimensionless; : Two different nodes in the network; : The current calculation time point; Through the above-mentioned refined calculations, the system not only considers the associations at the business level, but also incorporates topological factors at the physical and protocol layers such as switching behavior and signal coverage, thereby constructing a more comprehensive and accurate model of the coupling relationship between nodes.

[0023] Example 4: The handover impact weight and the signal quality impact weight are hyperparameters obtained through machine learning model training and optimization based on historical fault case data; The calculation of the fingerprint difference is specifically as follows: Calculating the Frobenius norm between the real-time coupling effect matrix and the coupling effect matrix of the prototype fingerprint at each time point within a preset fingerprint sequence length; and calculating the arithmetic mean of the Frobenius norms at all time points to obtain the fingerprint difference; This embodiment demonstrates the system's intelligent and adaptive capabilities. The handover impact weights and signal quality impact weights are not statically configured, but are instead optimized through machine learning models trained on a large amount of historical fault case data. This means the system can learn from real-world network operation and maintenance experience and automatically adjust parameters to achieve optimal fault location. At the same time, the calculation method of fingerprint difference is clearly defined. The system quantifies the distance between the real-time generated coupling effect matrix sequence and the prototype fingerprint in the fault library by calculating the Frobenius norm between them. This method is mathematically rigorous and can effectively measure the similarity between two high-dimensional matrix sequences. The specific calculation is as follows: ; Indicates the fingerprint difference; Represents the coupling effect matrix generated in real time; Represents the prototype fingerprint matrix corresponding to the kth fault type in the fault fingerprint library; Indicates the length of the fingerprint sequence used for comparison, which is a configuration parameter; represents the Frobenius norm, which is a standard matrix norm well known in the art and is used to calculate the difference between two matrices; In this way, the system transforms the complex fault location problem into a clear pattern matching problem, that is, finding the known fault mode that is most similar to the current network state.

[0024] Example 5: The fault root cause reverse reasoning module is also used for: Compare the calculated minimum fingerprint difference with a preset confidence threshold; When the minimum fingerprint difference is lower than the confidence threshold, outputting the positioning result; When the minimum fingerprint difference is not lower than the confidence threshold, outputting a report indicating that the fault type is unknown; The confidence threshold is determined by drawing a receiver operating characteristic curve on the validation dataset and based on preset positioning accuracy and recall rate targets; This embodiment illustrates the system's decision-making logic, ensuring the reliability of output results. After finding the prototype fingerprint with the highest matching degree, the fault root cause reverse reasoning module does not immediately output the result. Instead, it introduces a confidence threshold for secondary verification. Only when the minimum fingerprint difference falls below this threshold, meaning the similarity is sufficiently high, does the system confirm and output the location result. Conversely, if all known fault modes differ significantly from the current network status, the system outputs a "fault type unknown" report, avoiding misjudgments due to insufficient evidence. The scientific nature of this mechanism lies in the method for determining the confidence threshold. It is not set empirically, but rather scientifically determined by plotting the receiver operating characteristic (ROC) curve on a validation dataset. Operations and maintenance personnel can select the most appropriate balance point on the ROC curve as the threshold based on their varying requirements for positioning accuracy and recall. This data-driven decision-making mechanism ensures that the system's final output achieves a quantifiable and adjustable balance between reliability and coverage, greatly enhancing the system's practicality and robustness. Assume that in a 5G SA network consisting of multiple gNodeBs (base stations), the transmission optical cable of gNodeB-1 (gNB-1) experiences jitter, resulting in intermittent packet loss and increased latency for the user plane services it carries. Data collection and feature extraction: The network-wide flow topology data synchronization collection module collects data such as the decline in gNB-1 throughput, the increase in the number of active users of neighboring gNB-2 / gNB-3, and the increase in the number of handovers. The disturbance feature time series extraction module converts this raw data into feature time series such as the rate of change. ; Coupling effect quantification: The flow topology coupling effect quantification module receives the above-mentioned feature sequence and original topology data; calculate That is, between gNB-1 and gNB-2: the module calculates The high value is due to the strong latency correlation between the decrease in gNB-1 throughput and the increase in the number of users on gNB-2. The value is also high because of the number of handovers from gNB-1 to gNB-2 surge; final The result is a significantly high value; Similarly, calculate It is also a significantly high value; Since there is no direct strong causal relationship between gNB-2 and gNB-3, the calculated Relatively low; Finally, the generated coupling effect matrix In the figure, the elements related to gNB-1, i.e., the first row and the first column, show high values of mutation, forming a “star-shaped” divergent fingerprint pattern centered on gNB-1; Reverse reasoning and positioning: The fault root location module receives this "star-shaped" fault fingerprint sequence ; The module compares it with the fault fingerprint library and finds that it matches the prototype fingerprint marked as "single base station transmission failure" in the library. The difference D is the smallest and is less than the confidence threshold ; The system ultimately outputs the diagnostic conclusion: "Fault Root Location: gNB-1; Fault Type: Suspected Transmission Link Fault (High Confidence)." Based on this information, O&M personnel can directly troubleshoot the transmission link on gNB-1, significantly reducing troubleshooting time. The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. An AI-based fault routing rapid location system, characterized in that: include: The whole network flow topology data synchronization collection module is used to synchronously collect service flow indicator data and network topology indicator data from network elements in the wireless communication network to obtain synchronized original data; The disturbance feature temporal extraction module is used to calculate the time change rate and short-term volatility of service flow indicator data based on the synchronized raw data to generate a characteristic time series that represents the network status disturbance; a flow topology coupling effect quantification module, configured to combine the characteristic time series with the network topology indicator data to quantify the mutual influence strength between pairs of nodes in the network, so as to generate a real-time fault fingerprint sequence characterizing the mutual influence strength; A fault root cause reverse reasoning module is used to obtain a preset fault fingerprint library, which contains prototype fingerprints corresponding to multiple known fault types; And calculate the fingerprint difference between the real-time fault fingerprint sequence and each prototype fingerprint in the fault fingerprint library; The prototype fingerprint with the smallest fingerprint difference is determined, and the fault type and root node corresponding to the prototype fingerprint with the smallest fingerprint difference are used as the positioning result to generate a fault positioning report.

2. The AI-based fault routing rapid location system according to claim 1 is characterized in that: The disturbance feature temporal extraction module is specifically used for: Performing first-order difference calculation on the throughput time series in the business flow indicator data to obtain a throughput change rate series; For the time series of the number of active users in the business flow indicator data, calculate the standard deviation of the sliding window to obtain the volatility series of the number of active users; The throughput change rate series and the active user number volatility series are combined to generate the characteristic time series.

3. The AI-based fault routing rapid location system according to claim 1 is characterized in that: The flow topology coupling effect quantification module is specifically used for: Calculating the normalized flow disturbance correlation of the node pair; Calculating the dynamic topology impact factor of the node pair; Performing a weighted summation on the normalized flow disturbance correlation and the dynamic topology impact factor to obtain a flow topology coupling index; A coupling effect matrix is constructed based on the flow topology coupling index to generate the real-time fault fingerprint sequence.

4. The AI-based fault routing rapid location system according to claim 3 is characterized in that: The calculation of the normalized flow disturbance correlation is specifically as follows: Sliding the characteristic time series of the node pair within a preset time delay range, and calculating the Pearson correlation coefficient between the two series for each delay time point; The maximum absolute value of the correlation coefficient among all the delayed time points is selected as the normalized flow disturbance correlation.

5. The AI-based fault routing rapid location system according to claim 3 is characterized in that: The calculation of the dynamic topology impact factor is specifically as follows: Obtaining the number of successful handovers between the node pair in the current time window and dividing it by the historical average number of handovers to obtain a normalized handover behavior factor; Obtaining a reference signal received power from the source node reported by the user equipment of the target node, and subtracting a preset signal quality reference threshold to obtain a signal quality difference factor; multiplying the normalized handover behavior factor by a preset handover impact weight, and multiplying the signal quality difference factor by a preset signal quality impact weight; The two products are nonlinearly fused through a hyperbolic tangent function to generate the dynamic topology influence factor.

6. The AI-based fault routing rapid location system according to claim 5, characterized in that: The switching impact weight and the signal quality impact weight are hyperparameters obtained through machine learning model training and optimization based on historical fault case data.

7. The AI-based fault routing rapid location system according to claim 1, characterized in that: The calculation of the fingerprint difference is specifically as follows: Calculating the Frobenius norm between the real-time coupling effect matrix and the coupling effect matrix of the prototype fingerprint at each time point within a preset fingerprint sequence length; The arithmetic mean of the Frobenius norms at all time points is calculated to obtain the fingerprint difference.

8. The AI-based fault routing rapid location system according to claim 1 is characterized in that: The fault root cause reverse reasoning module is also used for: Compare the calculated minimum fingerprint difference with a preset confidence threshold; When the minimum fingerprint difference is lower than the confidence threshold, outputting the positioning result; When the minimum fingerprint difference is not lower than the confidence threshold, a report indicating that the fault type is unknown is output.

9. The AI-based fault routing rapid location system according to claim 8, characterized in that: The confidence threshold is determined by drawing a receiver operating characteristic curve on the validation dataset and based on preset positioning accuracy and recall rate targets.

Images