Smart city network flow prediction method and system

Abstract

The invention provides a smart city network traffic prediction method and system. The method comprises the following steps: preprocessing original sequential network traffic data based on a Kalman filter; based on a time sequence convolutional network, obtaining a target feature sequence of the preprocessed original time sequence network flow data; and inputting the target feature sequence into an LSTM network to obtain a prediction result of the original time sequence network flow data. The system executes the method. According to the invention, the Kalman filter, the TCN and the LSTM network are cooperatively combined together, noise existing in original sequential network flow data is eliminated by determining the Kalman filter, interference of the noise is reduced, high-precision prediction is achieved; then potential features are extracted from the data by adopting the TCN, and finally, the LSTM is adopted to predict the original time sequence network flow data.

Description

technical field [0001] The invention relates to the technical field of digital signal processing, in particular to a smart city network traffic prediction method and system. Background technique [0002] With the continuous advancement of information technology, smart cities have become a key construction task at present. In the various facilities of smart cities, the network is the most important thing. The huge urban infrastructure will bring a huge amount of network traffic data. Therefore, it is imperative to deal with network traffic and make high-precision predictions. An accurate and real-time network traffic prediction method can not only help network operators to better allocate resources and ensure service quality, but also detect malicious attacks on the network by comparing real traffic with predicted traffic. Basically, network traffic forecasting is a time series forecasting problem, and most traditional methods are designed based on statistical linear forecas...

AI Technical Summary

Problems solved by technology

[0003]However, with the continuous expansion of network scale and the rapid emergence of new services, network traffic often presents more unstable characteristics represented by strong nonlinearity

The oversimplified linear theoretical assumptions of traditional forecasting methods cannot effectively capture the nonlinear characteristics of time series in large-scale networks

[0004] In the prior art, a method based on the combination of Graph Neural Networks (GNN) and Long Short Term Memory networks (Long Short Term Memory networks, LSTM) is used. There is no Process the dataset to remove noise

Noise from this data will be very noticeable and may affect the processing of the original data, which may affect the accuracy of network traffic predictions

[0005]Based on the LSTM method to predict network traffic, to a certain extent, network traffic data can be predicted, but in terms of processing raw data and further extraction of data features It is difficult to further improve the accuracy of network traffic forecasting

Images