Research And Implementation Of Real-time Energy Monitoring System Based On Data Miming

With the increasing scale of the data center,the energy consumption comes grow exponentially.Data centers are faced with some problems,such as lower energy consumption,saving money,improving resource utilization,etc.Generally speaking,monitoring energy data to reduce the runtime energy consumption,which is an important way of energy saving.While in the traditional energy consumption monitoring system,the managerial personnel judge historical energy consumption whether is abnormal relying on setting threshold manually.The setting of threshold mainly depends on our subjective judgment,which lead to the lack of comprehensive,more accurate data to support the measures.On the other hand,more and more real-time energy consumption data is stored in the energy management platform.The historical energy consumption data often contains a wealth of knowledge,while conventional methods cannot effectively analyze massive amounts of data.However,Data mining methods have achieved good application effect in many fields.Therefore,how to apply the data mining method to the research on energy consumption data monitoring system has very important practical significance.In this context,this paper mainly completed the following work:First of all,analyzing the application scenarios of data mining algorithm in energy consumption system.Combined with the goal of energy consumption of abnormal detection,classification algorithm and outliers detection algorithm are mainly studied,and a real-time energy consumption data anomaly detection method is proposed.Then a large number of experiments are made to verify the validity of the method.Secondly,putting forward a new architecture for energy consumption monitoring system.By adding the middle server,request/IO and service logic are separated so that the service server can focus on dealing with the service logic,which reduce the load of service server effectively.The new architecture has good scalability,for example,with the increasing scale of service,we can add new service servers dynamically.Then,designing and realizing a energy consumption data real-time monitoring system.In the system,users can observe real-time energy consumption data with the help of many charts,and the abnormal detection methods are integrated and applied to the warning module of the system.The system can help administrators to monitor the energy consumption of the data center and inform administrators to view the abnormal situation of energy consumption.Finally,the specific function is showed and the performance test is made.Test results show that based on the new framework,the energy consumption real-time monitoring system shows good performance in high concurrency scenarios.