Research on Performance Diagnosis Method of Power User Eleco Energy Data Acquire System Based on Semi-Supervised Learning

摘要

With the overall coverage of power users' electricity information collection and the continuous growth of mining business, higher requirements are placed on the stability and reliability of the system. In order to solve the problems such as passive and incomplete in manual monitoring and diagnosis, an intelligent diagnosis method based on semi-supervised learning is proposed. Technologies such as Ping/Traceroute, SNMP protocol, SQL scripts and message queues are used to monitor and collect performance information of power-using information mining system, as well as information of related hardware devices and middleware in real time. Then store all kinds of monitoring information in a distributed manner. Finally, a performance diagnostic model of the system is constructed based on the semi-supervised learning algorithm, which can achieve intelligent diagnosis and early warning of system abnormalities. Experimental results show that this method can accurately diagnose system anomalies and provide technical support for quick elimination of failures and elimination of equipment hazards.