CN-122028105-A - Communication network performance evaluation method based on artificial intelligence

CN122028105ACN 122028105 ACN122028105 ACN 122028105ACN-122028105-A

Abstract

The invention relates to the technical field of communication networks, and particularly discloses a communication network performance evaluation method based on artificial intelligence, which comprises the steps of obtaining network historical operation data, setting performance critical judgment conditions, judging the current network critical balance state and marking a performance avalanche warning period, analyzing delay jitter and ping-pong switching at the starting moment of the warning period, calculating an evaluation value and judging whether to trigger performance avalanche warning, identifying a delay abnormal network point and a ping-pong switching network point after triggering the warning, constructing a new abnormal node identification model by combining a convolutional neural network and outputting a structural result, constructing a performance coupling causal graph by directional causal analysis, screening effective new abnormal network points, calculating the quantitative risk of the avalanche evaluation value, realizing the accurate positioning of network performance abnormality from an index level to a node level, capturing avalanche precursor characteristics, improving the warning accuracy and effectively evaluating the avalanche risk of the communication network performance.

Inventors

ZHU TAO
HE PENG
ZHOU FENG

Assignees

深圳建广数字科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260310

Claims (10)

1. The communication network performance evaluation method based on artificial intelligence is characterized by comprising the following steps: acquiring historical operation data of a communication network, setting a critical judgment condition of the performance of the communication network, judging the critical balance state of the current network, and marking a performance avalanche warning period according to a judgment result; Based on the performance avalanche warning period, carrying out communication network delay jitter and ping-pong switching analysis at the starting moment of the period, calculating a delay fluctuation evaluation value and a ping-pong switching evaluation value, and carrying out performance avalanche early warning triggering judgment; If triggered, based on the analysis results of the communication network delay jitter and ping-pong switching, identifying the delay abnormal network point and the ping-pong switching network point, constructing a new abnormal network point identification model, and outputting a structured new abnormal network point; Based on the structured new abnormal network points, constructing a performance coupling causal graph, and carrying out rationality screening on the structured new abnormal network points to determine effective new abnormal network points; And based on the effective new abnormal network point, combining the time delay abnormal network point and the ping-pong switching network point, and calculating an avalanche evaluation value of the current network performance.
2. The communication network performance evaluation method based on artificial intelligence according to claim 1, wherein the process of setting a communication network performance critical decision condition and making a current network critical balance state decision is as follows: The historical operation data of the communication network comprises transmission delay class data of data packets of each detection point, switching log class data of terminal equipment, load data of network nodes, bandwidth utilization rate data and packet loss rate data; Extracting transmission delay class data, load data, bandwidth utilization rate data and packet loss rate data of each detection point data packet; calculating the difference value duty ratio of the data and the corresponding critical standard based on any kind of data; if the difference value duty ratio is smaller than the critical standard, marking the data as critical performance data; Counting the number of categories of the critical performance data; When the number of categories of the critical performance data is more than or equal to 2, judging that the current network is in a critical balance state; and when the number of the categories of the critical performance data is less than 2, judging that the current network is not in a critical balance state.
3. The method for evaluating performance of a communication network based on artificial intelligence according to claim 1, wherein the step of marking the performance avalanche warning period according to the judgment result is as follows: If the current network is in a critical equilibrium state: extracting all critical balance state duration time in the historical operation data of the communication network; Calculating the average value of all the critical balance state duration time in the historical operation data of the communication network to obtain the estimated duration time of the current critical balance state; Backtracking real-time operation data from the current moment, and judging moment by moment according to performance critical judgment conditions; determining the starting time of the current critical balance state, and subtracting the starting time from the current time to obtain the duration of the current critical balance state; performing difference operation on the estimated duration and the sustained duration to obtain the current residual critical duration; The time period from the current moment to the end moment of the current residual critical duration is marked as a performance avalanche warning period.
4. The communication network performance evaluation method based on artificial intelligence according to claim 1, wherein the communication network delay jitter and ping-pong handover analysis is performed at the beginning of the period, and the process of calculating the delay fluctuation evaluation value and the ping-pong handover evaluation value is as follows: based on historical operation data of the communication network, extracting the transmission delay of each detection point data packet in the current critical duration time, and integrating the transmission delay into a transmission delay sequence; Calculating the variation coefficient of the transmission delay sequence to obtain a delay fluctuation evaluation value; extracting a switching log of the terminal equipment within the current critical duration; And counting the occurrence times of ping-pong switching to obtain a ping-pong switching evaluation value.
5. The communication network performance evaluation method based on artificial intelligence according to claim 1, wherein the performance avalanche warning triggering judgment process is as follows: Setting early warning judgment conditions: The delay judgment condition is that the delay fluctuation estimated value exceeds the delay jitter abnormal standard, and the average value of the transmission delay sequence is in the normal delay interval; Switching judgment conditions that the ping-pong switching evaluation value exceeds the ping-pong switching frequency standard and the repeated switching behavior of the terminal among the nodes is not recognized currently; The repeated switching behavior means that the terminal only switches back and forth between two nodes, and the bidirectional switching times reach the ping-pong switching frequency standard; When the time delay judgment condition and the switching judgment condition are met simultaneously, the communication network is judged to enter a performance avalanche degradation stage, and the performance avalanche early warning is triggered immediately.
6. The method for evaluating the performance of a communication network based on artificial intelligence according to claim 5, wherein the step of identifying the network point with abnormal delay and the network point for ping-pong handover comprises the following steps: The transmission delay of each detection point data packet in the current critical duration time is associated with the attributive network node one by one; marking the network nodes meeting the time delay judgment condition as time delay abnormal network points; counting the total number of times of each network node participating in ping-pong switching in unit time to obtain a ping-pong switching evaluation value of the network node; and if the ping-pong switching evaluation value is greater than or equal to the ping-pong switching frequency standard of the node level, marking the network node as a ping-pong switching network point.
7. The communication network performance evaluation method based on artificial intelligence according to claim 1, wherein the process of constructing a new abnormal network point identification model and outputting a structured new abnormal network point is as follows: the new abnormal network point identification model comprises an input layer, a convolution layer, a pooling layer and a full connection layer; characteristic data of the delay abnormal network point and the ping-pong switching network point are taken as model input; The characteristic data refer to a time delay abnormal network point, a node identifier of a ping-pong switching network point, a time delay fluctuation evaluation value and a ping-pong switching evaluation value; Extracting topological association and abnormal diffusion characteristics through a convolution layer; performing dimension reduction processing on the extracted topological association and abnormal diffusion characteristics through a pooling layer; And the full connection layer analyzes the abnormal development trend and outputs a structured new abnormal network point.
8. The method for evaluating performance of an artificial intelligence based communication network according to claim 5, wherein the process of constructing the performance coupling causal graph comprises the steps of: The method comprises the steps of obtaining surrounding network nodes associated with a structured new abnormal network point, and taking the surrounding network nodes and the structured new abnormal network point as node entities of a causal graph; taking the delay fluctuation evaluation value and the ping-pong switching evaluation value of the detection point data packet as index entities; carrying out causal inspection by adopting a Grangel causal inspection, judging the abnormal propagation direction, and marking causal direction for the node entity; If the network node only meets the time delay judgment condition, judging that the abnormality is caused by the time delay abnormality, if the network node only meets the switching judgment condition, judging that the ping-pong switching is abnormal, and if the network node simultaneously meets the early warning judgment condition, judging that the abnormality is caused by the mixing abnormality; The cause of the abnormality is taken as an abnormality type; and integrating the node entity, the index entity, the causal direction and the anomaly type to obtain a performance coupling causal graph.
9. The communication network performance evaluation method based on artificial intelligence according to claim 1, wherein the process of performing rationality screening on the structured new abnormal network points and determining effective new abnormal network points is as follows: the rationality screening comprises causal association screening, abnormal cause matching screening and conduction feasibility screening; causal relevance screening, namely, causal pointing exists in a node entity; Conducting feasibility screening, namely enabling node entities to be consistent with index entities of which causal points to corresponding node entities; the abnormal incentive matching screening, wherein the node entity and the cause and effect of the node entity are the same as the abnormal incentive pointed to the corresponding node entity; and marking the network nodes which pass the three times of screening as effective new abnormal network points.
10. The method for evaluating performance of a communication network based on artificial intelligence according to claim 1, wherein the step of calculating the avalanche evaluation value of the current network performance is: Taking the union of the delay abnormal network point and the ping-pong switching network point as an abnormal node set; Counting the number of overlapped nodes of the effective new abnormal network point which is the same as the number of overlapped nodes of the abnormal node set, and the total number of nodes of the effective new abnormal network point; and taking the ratio of the number of overlapped nodes to the total number of nodes of the effective new abnormal network point as an avalanche evaluation value of the current network performance.

Description

Communication network performance evaluation method based on artificial intelligence Technical Field The invention relates to the technical field of communication networks, in particular to a communication network performance evaluation method based on artificial intelligence. Background With the rapid development of the 5G and Internet of things technologies, the scale of a communication network is continuously enlarged, node connection is more and more dense, stable operation of network performance becomes a core for guaranteeing smooth development of various services, and the communication network performance evaluation technology based on artificial intelligence has become an important direction of industry development because of the advantages of high efficiency and accuracy. However, the existing communication network performance evaluation method still has obvious defects, is difficult to adapt to performance monitoring requirements under complex network scenes, particularly cannot meet practical application requirements in high-density node deployment, 5G antenna deployment, multi-service concurrent transmission and terminal direct-connection scenes, and has the two particular problems that firstly, the existing evaluation method does not fully utilize a machine learning technology to mine node association relations, dynamic abnormal behaviors such as ping-pong switching, abnormal diffusion and the like of a network node in a terminal access process are not accurately identified, fine characteristics of transition from a network critical state to an avalanche stage cannot be captured, early warning is difficult to trigger in advance to avoid performance deterioration, secondly, network topology association data are not fully combined in an evaluation process, performance coupling relations and abnormal conduction paths among nodes are not considered to be insufficient, deep analysis on causal association of the nodes is lacked, network abnormal conduction risks are difficult to be determined, and reliability of evaluation results is further influenced. Therefore, the invention provides a communication network performance evaluation method based on artificial intelligence. Disclosure of Invention The invention aims to provide a communication network performance evaluation method based on artificial intelligence so as to solve the problems in the background. The aim of the invention can be achieved by the following technical scheme: A communication network performance assessment method based on artificial intelligence, comprising: acquiring historical operation data of a communication network, setting a critical judgment condition of the performance of the communication network, judging the critical balance state of the current network, and marking a performance avalanche warning period according to a judgment result; Based on the performance avalanche warning period, carrying out communication network delay jitter and ping-pong switching analysis at the starting moment of the period, calculating a delay fluctuation evaluation value and a ping-pong switching evaluation value, and carrying out performance avalanche early warning triggering judgment; If triggered, based on the analysis results of the communication network delay jitter and ping-pong switching, identifying the delay abnormal network point and the ping-pong switching network point, constructing a new abnormal network point identification model, and outputting a structured new abnormal network point; Based on the structured new abnormal network points, constructing a performance coupling causal graph, and carrying out rationality screening on the structured new abnormal network points to determine effective new abnormal network points; And based on the effective new abnormal network point, combining the time delay abnormal network point and the ping-pong switching network point, and calculating an avalanche evaluation value of the current network performance. Further, setting a critical decision condition of the performance of the communication network, and performing the current critical balance state decision of the network comprises the following steps: The historical operation data of the communication network comprises transmission delay class data of data packets of each detection point, switching log class data of terminal equipment, load data of network nodes, bandwidth utilization rate data and packet loss rate data; Extracting transmission delay class data, load data, bandwidth utilization rate data and packet loss rate data of each detection point data packet; calculating the difference value duty ratio of the data and the corresponding critical standard based on any kind of data; if the difference value duty ratio is smaller than the critical standard, marking the data as critical performance data; Counting the number of categories of the critical performance data; When the number of categories of the critical performance data i