Research Of Data Analysis And Intelligent Fault Detection On Wireless Battery Monitoring Systems

VRLA batteries, the UPS energy storage devices, are widely used as the backup power system in telecommunication, power, transportation and other fields. Their performance directly determines the reliability of the whole system. If used improperly, individual cells often ran out early. Therefore, it is of great significance to monitor the working statues of the batteries on real time.The battery monitoring system used in this paper is based on wireless networking technology and achieves a real-time online monitoring of some main parameters, including the terminal voltage of each battery, the battery shell’s temperature, the battery bank voltage, working current, ambient temperature. The system uses wireless communication instead of cable transmission, and establishes a monitoring network and has the alarm management part. The main work of this thesis is to improve the detection accuracy of each parameter, and achieve real-time diagnosis of various types of faults of batteries based on the current and historical data, and provide all kinds of alarms. The main researches this paper has completed are as follows:(1) A cross-network repeater is designed to construct a multilevel ZigBee network. It is aimed to solve the monitoring problem in metal sealed environment and makes ZigBee networking construction and signal transmission possible in such environment. VRLA batteries, the monitoring object, are stored in a UPS battery cabinet which is usually metallic materials. The design solves the monitoring problem of battery in this environment.(2) Improving the measurement accuracy of the system parameters. The accuracy of terminal voltages reaches±10mV through the software calibration, and the currents’ accuracy reaches±50mA by introducing the method of eliminating the zero drift. And the terminal nodes can simultaneously measure the temperature of the battery without the effect of PCB fever via mold design.(3) Proposing a method of data analysis and fault detection of batteries and dividing batteries’fault into four categories:linkage fault, charge-discharge fault, inconsistency and aging of batteries. This paper determines the system fault of linkage and charge-discharge through the real-time data, builds historical data archives and estimates inconsistency and health of batteries through the process of charging-discharging and the historical data archives. The paper proposes three basic parameters for judging the performance of the battery:time to fully discharge, steep drop in voltage and the fluctuation of discharge voltage; And the experiments verifies the feasibility of fault judgment through these three parameters extracted from the batteries discharge curve. The fault detection scheme presented in the paper does not require the measuring of batteries internal resistance and significantly saves the system cost.