Big Data Based Online Prediction Of Battery State With Uncertainty Analysis

So far,China government has taken the new energy car as the strategic emerging industry,which gets rapid development due to various forces.However,some key problems remain to be solved because of that we are not capable of breaking through in key technical fields.As we all know,the lithium-ion battery has birth defects: instability,short-life,unreliability,and insecurity,which limit the application and promotion of the new energy cars.Firstly,the importance and necessity of big data method for analysis of battery state is explained.The core value of “Big Data” is its ability to store and analyze the massive data sets,which make it an optimization in the composite cost of price,speed and ability.And the new energy car monitoring center can collect large volume of real-time data sets of operating conditions.Then we list different kinds of research methods in prediction and analysis of traction battery's real-time state.And the data set management of SOH is carried out in Matlab after getting rid of the abnormal data based on the powerful features of storage and collection of the monitoring center.Secondly,combining with the principle of phase-space reconstruction,the change of SOH,which forms the one-dimensional phase space can be handled as the chaotic system.And the minimum embedded dimension and the best delay time is determined to restore dynamic characteristic of SOH,from which the iterative equation is derived.Thirdly,the principle of RVM algorithm is introduced,which contains the formula derivation and drawbacks.Then the improved algorithm is carried out to decrease the computation complexity and increase the computation speed.And iterative equation is utilized to calculate the SOH to verify the accuracy and speed.The new QMU verification method guarantees the reliability of numerical analysis,which fits the online prediction better.Finally,taking the random factors of SOC into consideration contributes to extraction of substantive characteristics,we divide the route into 28 segments.And every segment is approximated as uniform acceleration,uniform velocity or uniform deceleration.The histogram is plotted to reflect the relationship between mean SOC and acceleration.Thereby,we form an evaluation system for the dynamics of traction battery's SOC with uncertainty analysis,which provides the numerical interval analysis with top and bottom limitation.