Design And Research Of Fault Diagnosis System On Vanadium Redox Flow Battery

Efficient large-scale energy storage battery is the core of the new energy industry revolution,photovoltaic,wind power and other new energy inherent randomness,volatility,intermittent,peak load characteristics,determines its large-scale development must have advanced energy storage technology for support.All-vanadium flow battery(VRB)is the largest,most technologically advanced and close to industrialized high-performance rechargeable battery in the world.It has high power,high energy,high efficiency,low cost,long life and high safety.no pollution and a series of outstanding advantages.It has a very good application prospect in the fields of photovoltaic power generation,wind power generation,smart grid,distributed power station,communication base station,emergency power supply,traffic municipal,military power storage and so on.In the VRB energy storage system,VRB is a key component of the energy storage system,but also the main source of failure of the energy storage system.The failure research of VRB system can effectively diagnose VRB system fault and predict the occurrence of VRB system fault.It has great engineering application value to improve the service life of VRB.This paper analyzes the working principle of VRB,and builds the VRB equivalent loss model based on the working principle in simulink simulation platform,and then identifies the theoretical basis for the later.And the VRB online SOC estimation scheme based on EKF(extended Kalman filter)algorithm is designed.Secondly,it explains the importance of the VRB group’s equalization to VRB inconsistency faults.Based on the equilibrium and mean method of SOC as the equilibrium strategy,an equalization control scheme of vanadium battery pack based on Fischer capacitance method is designed.Finally,a vanadium battery fault diagnosis and detection module is designed and the vanadium battery is diagnosed by fuzzy BP neural network.The reliability of the trained network is verified by four sets of typical fault samples.