CN-121984835-A - Intelligent research and judgment and artificial intelligent auxiliary emergency treatment method for communication faults

CN121984835ACN 121984835 ACN121984835 ACN 121984835ACN-121984835-A

Abstract

The invention discloses an intelligent studying and judging and artificial intelligent auxiliary emergency treatment method for communication faults, which relates to the technical field of communication fault treatment and aims to solve the technical problem that secondary faults are easy to occur due to lack of a scheme verification mechanism in the conventional communication fault treatment at present. According to the invention, the network real-time state under the current fault scene can be completely re-carved by establishing the high-fidelity network digital twin body, and the network state change after each candidate plan is executed can be simulated by combining with a network state equation, so that the finally executed plan is ensured to have the effectiveness and the safety, and the problem that the secondary fault is easy to be caused due to the lack of a plan verification mechanism in the current traditional communication fault treatment is solved.

Inventors

GENG SHAOBO
SU HAN
ZHANG JIAJU
LI CHAO
SU MENG
MIAO YONG
DONG ZHIMIN

Assignees

国网河北省电力有限公司

Dates

Publication Date: 20260505
Application Date: 20260131

Claims (10)

1. The intelligent research and judgment and artificial intelligent auxiliary emergency treatment method for the communication faults is characterized by comprising the following steps of: s100, data perception and feature extraction, namely constructing a real-time data mirror image of a communication network through multi-source data real-time acquisition, data preprocessing and feature engineering; S200, intelligent research and judgment, namely adopting an isolated forest algorithm to detect and identify faults and adopting an intelligent research and judgment core engine based on a graph propagation algorithm and a Bayesian network to realize quick accurate qualitative, delimitation and grading of the faults; s300, AI auxiliary decision making and plan generation, namely forming an optimal disposal plan through intelligent matching of a historical plan, digital twin simulation deduction and verification and multi-objective optimization decision making and plan generation; s301, combining a historical fault case library, calculating the similarity between vectors by using the characteristic of the current fault and adopting a cosine similarity algorithm, quickly finding out the most similar past case, and taking a successful treatment scheme as an important reference; S302, establishing a high-fidelity network digital twin body, and performing simulation by repeatedly engraving a current fault scene and a network real-time state in the twin body and then performing simulation by performing simulation according to the operation instruction of each candidate plan one by one and according to a time sequence to form an optimal treatment scheme; s303, based on quantized data obtained by simulation deduction, the system uses a TOPSIS multi-attribute decision algorithm to carry out scientific sorting, and a recommended treatment plan is generated; s400, treatment execution and effect evaluation, namely realizing safe and effective execution of a treatment scheme through human-computer cooperative treatment, plan execution and automatic repair and treatment effect real-time evaluation; s500, knowledge base and model self-learning, namely realizing continuous self-evolution of the system by constructing a fault case knowledge base and a model optimization mechanism.
2. The method for intelligent studying and judging of communication failure and artificial intelligence auxiliary emergency treatment according to claim 1, wherein in the step S301, the cosine similarity algorithm calculation formula is as follows: ; Wherein, the As the dot product of vectors a and B, 、 Respectively is vector 、 Euclidean norms, similarity of (d) The value range is [0,1], and the closer the value is to 1, the higher the matching degree of the root cause characteristics of the current fault and the historical case is represented.
3. The method for intelligent research and judgment of communication failure and artificial intelligence assisted emergency treatment according to claim 1, wherein in step S302, the network state equation is as follows: , , wherein, For a link Is used for the packet loss rate of the mobile terminal, For a link Is used for the time delay of (1), For a link Is provided).
4. The method for intelligent studying and judging of communication failure and artificial intelligence assisting emergency disposal according to claim 1, wherein the step S303 further comprises the steps of: s3031, constructing decision matrix, defining matrix : Wherein Representing the number of candidate protocols, Representing the number of evaluation indexes, such as recovery time, business impact surface, and operation risk coefficient, Represent the first The proposal is at the first Quantized values under the individual indicators; S3032, normalizing the matrix, namely carrying out normalization processing on the decision matrix by adopting an Euclidean distance method to eliminate the influence of the difference of different index dimensions, wherein the calculation formula is as follows: , wherein, Is a matrix element after normalization, Represent the first The proposal is at the first Quantized values under the individual indicators; s3033, ideal and understood Is the optimal value set under each index dimension, such as the index value with the shortest recovery time and the smallest influence surface, and the negative ideal solution The index value is the worst value set under each index dimension, such as the index value with the longest recovery time and the largest influence surface; s3034, calculating Euclidean distance, namely respectively calculating Euclidean distance from each plan to positive ideal solution and negative ideal solution: Plan of To the right ideal Is a distance of (2); Plan of To negative ideal solution Is a distance of (2); s3035, calculating the relative closeness by a formula Calculating the relative closeness of each plan; Wherein, the The value range of (1) is [0 ], the closer the value is to 1, the closer the scheme is to the optimal scheme, and accordingly the priority ranking of all candidate schemes is completed.
5. The method for intelligent studying and judging of communication failure and artificial intelligence assisting emergency treatment according to claim 1, wherein the step S100 further comprises the steps of: S101, multi-source data real-time acquisition, namely acquiring data in parallel through a probe, an agent and a standard interface, wherein the data comprises equipment and link alarm data, service KQI, network KPI, equipment operation log, signaling tracking data and machine room dynamic ring infrastructure data, and all the data are converged through a high throughput message middleware; s102, data preprocessing and feature engineering, namely extracting feature vectors capable of deeply reflecting the essence of the network state from original data by adopting a dynamic threshold algorithm, and providing effective input for a subsequent AI model; the calculation formula of the dynamic threshold algorithm is , wherein, Is at the point of time Is used to determine the predicted value of (c), Is the standard deviation of the prediction error and, Is the sensitivity coefficient, when the real-time data Beyond the limit of And triggering abnormal detection when the range is in the range, so that the service period change is more self-adaptive.
6. The method for intelligent studying and judging of communication failure and artificial intelligence assisting emergency treatment according to claim 5, wherein the step S200 further comprises the steps of: S201, fault detection and identification, wherein the system adopts a dynamic threshold algorithm and an isolated forest algorithm to analyze the inflow characteristic data flow in real time, and abnormal data points are obtained Its anomaly score The approximation can be: ; Wherein, the Is a data point The desire for path length in a plurality of orphaned trees, Is a given number of samples Normalization factor when When the value is close to 1, the determination is made Is an outlier.
7. The method for intelligent studying and judging of communication failure and artificial intelligence assisting emergency disposal according to claim 1, wherein the step S200 further comprises the steps of: And S202, an intelligent research and judgment core engine is used for realizing quick qualitative, accurate delimitation, clear root cause and scientific grading of faults through the deep analysis after fault detection and confirmation on the basis of data perception and feature extraction in the step S100, and providing a key basis for subsequent auxiliary decisions.
8. The method for intelligent studying and judging of communication failure and artificial intelligence assisting emergency disposal according to claim 7, wherein the step S202 further comprises the steps of: S2021, fault delimitation and positioning, namely analyzing a propagation path of a fault signal by using a graph propagation algorithm, wherein the fault is transmitted from a source node Initially, it propagates to nodes Is used for the failure influence of the (c) in the (c), The calculation can be iterated by the following formula: ; Wherein, the Is a node Is used to determine the neighbor set of a neighbor, Is a node Is used for the degree of (a), Is a damping factor that represents the probability of the fault continuing to propagate, Is when Is an indication function with a source fault point value of 1; S2022, root cause analysis, wherein on the basis of positioning, an engine combines equipment logs, configuration change records and performance baselines, uses a Bayesian network interpretable AI model to carry out probabilistic reasoning, and gives an observed fault phenomenon Such as multiple alarm events, the most likely root cause Is the one that maximizes the posterior probability: ; Wherein, the Is the root cause The prior probability from the historical data is derived, Is at root cause Under the occurrence conditions, a phenomenon is observed Likelihood probability of (2); And S2023, evaluating the influence range and the grade, wherein the engine automatically analyzes the influenced clients, the influenced service types and the influenced number according to the service resource mapping table, and automatically evaluates the grade of the fault according to the breadth, the depth and the service importance of the comprehensive fault and a preset service level agreement SLA rule so as to provide a basis for the treatment priority.
9. The method for intelligent studying and judging of communication failure and artificial intelligence auxiliary emergency treatment according to claim 1, wherein the step S400 further comprises the steps of: S401, man-machine co-processing, namely pushing the generated intelligent work order to operation and maintenance personnel, and providing clear decision basis by a system, such as root cause analysis conclusion and simulation contrast chart, and finally confirming, fine-tuning or scheme selection by the personnel; S402, executing the plan and automatically repairing, namely automatically executing the standardized and highly-repeated operation through a network controller after the operation is authorized, decomposing the complicated operation into steps to guide operation and maintenance personnel to execute, and providing secondary confirmation and error proofing verification; S403, evaluating the treatment effect in real time, namely automatically judging whether the service is restored to a normal level and kept stable by comparing index changes before and after treatment, and simultaneously monitoring whether a new abnormal alarm is generated by the system to evaluate whether side effects are introduced in the treatment operation and automatically generating a treatment effect report.
10. The method for intelligent research and judgment of communication faults and artificial intelligence assisted emergency treatment according to claim 1, wherein the model self-learning in the step S500 comprises optimizing a research and judgment model by adopting an incremental learning algorithm, adjusting a decision strategy through reinforcement learning, comparing simulation with an actual feedback optimized digital twin model, and storing full-link treatment data into a fault case knowledge base in a structured manner to realize continuous evolution of the system.

Description

Intelligent research and judgment and artificial intelligent auxiliary emergency treatment method for communication faults Technical Field The invention relates to the technical field of communication fault treatment, in particular to an intelligent research and judgment and artificial intelligent auxiliary emergency treatment method for communication faults. Background As an important infrastructure of modern society, the stability and reliability of communication networks are directly related to the normal operation of socioeconomic. With the rapid development of novel communication technologies such as 5G, internet of things and industrial Internet, the modern communication network is continuously enlarged in scale, the network structure is increasingly complex, the service types are increasingly diversified, and the factors bring unprecedented challenges to the operation and maintenance management of the communication network. In the conventional communication fault handling, the candidate plan is lack of an early verification mechanism after being generated, an operation and maintenance personnel can only judge the feasibility of the plan by experience and directly execute the plan on a physical network, the communication network is a highly-associated complex system, the operation of a certain device or link influences a plurality of nodes through topological association and service bearing relations, when a fault port is reset, if the multi-service mapping relation borne by the port is not mastered, the associated voice, data and video services can be interrupted at the same time, when a routing strategy is adjusted for repairing a fault in a certain area, if the link load after the traffic redistribution is not calculated, the adjacent link can be paralyzed due to traffic overflow, and because the network state changes such as link traffic load fluctuation, the influence of the device operation on the associated service after the execution of the plan cannot be prejudged, the secondary fault is easily caused by the direct execution, the malignant cycle for repairing the new fault is formed, the network stability is seriously influenced, and the requirements of the current communication network on the fault handling safety and reliability are difficult to meet. In view of this, we propose an intelligent studying and judging of communication faults and an artificial intelligence assisted emergency disposal method. Disclosure of Invention The invention aims to overcome the defects of the prior art, adapt to the actual needs, and provide an intelligent studying and judging and artificial intelligent auxiliary emergency treatment method for communication faults, so as to solve the technical problem that secondary faults are easy to cause due to lack of a plan verification mechanism in the conventional communication fault treatment. In order to solve the technical problems, the invention provides the following technical scheme that the intelligent research and judgment and artificial intelligent auxiliary emergency treatment method for communication faults comprises the following steps: s100, data perception and feature extraction, namely constructing a real-time data mirror image of a communication network through multi-source data real-time acquisition, data preprocessing and feature engineering; S200, intelligent research and judgment, namely adopting an isolated forest algorithm to detect and identify faults and adopting an intelligent research and judgment core engine based on a graph propagation algorithm and a Bayesian network to realize quick accurate qualitative, delimitation and grading of the faults; s300, AI auxiliary decision making and plan generation, namely forming an optimal disposal plan through intelligent matching of a historical plan, digital twin simulation deduction and verification and multi-objective optimization decision making and plan generation; s301, combining a historical fault case library, calculating the similarity between vectors by using the characteristic of the current fault and adopting a cosine similarity algorithm, quickly finding out the most similar past case, and taking a successful treatment scheme as an important reference; S302, establishing a high-fidelity network digital twin body, and performing simulation by repeatedly engraving a current fault scene and a network real-time state in the twin body and then performing simulation by performing simulation according to the operation instruction of each candidate plan one by one and according to a time sequence to form an optimal treatment scheme; s303, based on quantized data obtained by simulation deduction, the system uses a TOPSIS multi-attribute decision algorithm to carry out scientific sorting, and a recommended treatment plan is generated; s400, treatment execution and effect evaluation, namely realizing safe and effective execution of a treatment scheme through human-computer cooperative treatment, pla