Network equipment early warning prototype system

Abstract

A network equipment early warning prototype system comprises a machine room data acquisition device, an API interface I / O, a database memory, a time sequence fault prediction model and a fault prediction GUI interface which are electrically connected in sequence. The function of the time sequence fault prediction model is that a current DBN model based on an RBM algorithm is realized through a python language, and historical data is called for model training; for the trained model, the system outputs time series fault early warning data in a future period of time according to the real-time data of the existing network machine room; the function of the fault prediction GUI interface is as follows: through a system GUI interface developed by QT software, early warning data realized by background model operation is extracted from the whole interface, the time sequence early warning condition of each device is displayed in the form of a trend curve, and red light display is carried out ondevices with high-risk faults recently. The network core router management and control method is applied to network core router management and control in the smart campus operation and maintenance process.

Description

technical field [0001] The invention relates to a core router failure prediction method based on a deep learning algorithm, and proposes a prediction prototype system according to the method. The invention can be applied to the management and control of the network core router in the operation and maintenance process of the smart campus, so as to realize intelligent network equipment failure early warning. Background technique [0002] The existing fault time series methods mainly use statistical methods, including SSA algorithm, ARIMA algorithm, support vector regression machine, etc. The above algorithm combs through the historical fault data collection and applies the corresponding algorithm to generate the fault model. By applying the failure model in the analysis system, the failure prediction of the existing system is realized. [0003] The three algorithms mentioned in the existing fault time series can only predict a certain parameter in the sequence (for example, ...

AI Technical Summary

Problems solved by technology

[0005] 1. The construction process of different algorithms is complicated, relying on a large number of manual participation experiments;

[0006] 2. The system composed of different statistical algorithms has poor fit, and the generated model has a high degree of nonlinearity, which is not suitable for practical use

Images