Research And Application Of Forecast Algorithm Based On AIOps

With the advent and rapid development of the information age,the IT systems of countless companies such as the Internet,banks,and e-commerce companies are becoming larger and larger,and the required operation and maintenance costs are getting higher and higher.The traditional operation and maintenance architecture can no longer support the efficiency of the IT system.Work,so the intelligent operation and maintenance AIOps(Artificial Intelligence for IT Operations)came into being.AIOps combines big data and machine learning algorithms to achieve a low-cost,high-efficiency operation and maintenance model,giving major enterprises more room for development.AIOps has many business scenarios,such as anomaly detection,root cause analysis and prediction,among which AIOps prediction technology is very important,mainly divided into scenarios such as failure prediction,performance prediction,capacity prediction,and transaction volume prediction.Predictive analysis of transaction volume and capacity can help realize intelligent system capacity planning,which can not only expand capacity in time during peak business periods to ensure high-performance system operation,but also reduce resource allocation and reduce costs in other periods.Therefore,this article starts with the forecasting technology of AIOps,and mainly conducts in-depth research on the time series forecast of transaction volume and disk capacity in AIOps.First,the development of related technologies in intelligent operation and maintenance AIOps and its forecasting business scenarios and the current research status at home and abroad are studied;secondly,the relevant knowledge of intelligent operation and maintenance AIOps is introduced,mainly including the basic concepts and intelligentization of intelligent operation and maintenance.Operation and maintenance capability classification and framework,IT operation and maintenance development process and comparison of various stages of operation and maintenance mode,advantages of intelligent operation and maintenance and future development trends;again,introduces the advantages and disadvantages of time series forecasting basic algorithms and time series forecasting models The evaluation indicators of the company have studied the principles of time series forecasting models such as Prophet model,ARIMA model,SARIMA model,exponential smoothing model and LSTM model,and in-depth study of the principle of combined forecasting model and model combination methods,and aiming at transaction volume forecasting and disk capacity The forecast puts forward the DM-PS4DB combination forecasting model based on the advantage matrix method.Its core idea is to use the weighting method of the advantage matrix method to combine the prediction results of the Prophet model and the SARIMA model to obtain new prediction results,fully combining the advantages of the two,More accurate and stable forecast of transaction volume and capacity;finally,using the open source transaction volume and disk capacity historical sequence data set,applying each single forecast model and the proposed combination forecast model to carry out modeling and forecasting experiments respectively,according to the results of the evaluation indicators As a result,the prediction effect of the DM-PS4DB combined model is more accurate and stable than each single prediction model.It is hoped that the research in this article can provide ideas and play a certain role for the change of transaction volume prediction and disk capacity prediction technology in intelligent operation and maintenance AIOps.