基于AHP、ARIMA算法的电力信息系统负载预测研究与应用

The application of AHP and ARIMA in the Load Prediction of Power Information System

电网公司已建成信息系统数量较多,服务器承载的信息量和访问量呈几何级数增长,在这种情况下,由于长时间累积运行可能导致系统超负荷运行或者负荷过于空闲。针对此问题,从信息系统负载角度对系统进行状态评价,首先利用电网公司的大量历史数据指标建立指标体系,其次利用AHP方法综合CPU利用率、内存利用率、磁盘利用率3个指标得到综合资源利用率,最后利用ARIMA方法和holt-winters指数平滑方法分别对综合资源利用率和系统响应时长进行预测,根据预测结果进行分级预判系统负荷状态,评估业务系统资源是否需要进行调整,满足用户对系统的要求,并保证资源的有效利用,减少资源空闲和浪费,优化资源调度。

As the power grid has built a number of information systems,the amount information carried by the server and number of visits to the server is growing exponentially. In this case,the long-time and accumulated operation may lead to the overload or too idle operation of the system. With this regard,the system should be evaluated from the perspective of the information system load. In this paper,first,we use a large amount of historical data of grid companies to establish an index system,and then use the AHP method to integrate three indices of CPU utilization,memory utilization and disk utilization to obtain the overall resource utilization rate,and finally we forecast the overall resource utilization and system response time using ARIMA method and holt-winter exponential smoothing method respectively. According to the forecast results we grade and pre-judge the system load status and judge whether the business system resources need to be adjusted to meet the user＇s requirement for the system and ensure the efficient use of the resources and reduce the idle and waste of resources and optimize the resource scheduling.