A Host Load Forecasting Method Based on Multi-sequence Combination

Abstract

The invention relates to a host load prediction method based on multi-sequence combination, which belongs to the field of computer application technology. From the perspective of nonlinear and non-stationary information processing, the present invention constructs multiple sequences and utilizes various data relationships related to each sequence and load, and combines the wavelet-AR-SVR-MA model and the AR model to predict the host load. To improve the host load prediction accuracy. This method is suitable for the host load prediction of key periods or key moments.

Description

technical field [0001] The invention relates to a host load prediction method based on multi-sequence combination, which belongs to the technical field of computer applications. Background technique [0002] Generally, in a distributed network environment, load forecasting is divided into network-based load forecasting (network load forecasting) and host-based load forecasting (host load forecasting). The research object of network load forecasting is the load of a specific network, and network traffic is usually regarded as a load indicator for forecasting. The host load prediction is concerned with the load of a specific host. The indicators proposed in the early stage mainly include CPU performance indicators, disk and memory available space, and process response time. With the development of network technology, when a host in the network mainly provides network services, the network traffic of the host becomes an important indicator to measure its load. [0003] Host l...

AI Technical Summary

Problems solved by technology

[0019] However, the disadvantage of this method is that, for historical sequences, each branch after wavelet decomposition uses BP neural network and AR prediction, while the literature "Research on Chaotic Time Series Prediction Based on Support Vector Machine and Wavelet Theory" (University of Science and Technology of China Doctoral dissertation. 2003) clearly pointed out that BP neural network has obvious defects compared with the support vector machine (SVM) method, such as learning is easy to fall into local minimum points, and it is easy to cause over-fitting and affect the generalization ability of the network, etc.

The above defects become the main reason affecting the prediction accuracy