Knowledge-data Coupling Driven Method for Data Cleaning and Recovery of Cloud Database of Lithium Batteries on Electric Vehicles

As the increasing inventory of new energy vehicles, large amounts of operational data have been generated and uploaded to cloud platform. Unfortunately, the numerous databases usually consist of abnormal data or data loss, which indicating that it cannot be applied for modelling and algorithm directly. Thus, the data cleaning and recovery is necessary which gets rid of abnormal data and become consequent especially for guaranteeing the precision and robustness for algorithm. In this article, a knowledge-data coupling driven method is proposed for data cleaning and recovery method, where a coupling model is used for simulating lost data. Based on simulation results originated from experiments, a satisfactory consistency is validated with more than 95% precision for recovery. The proposed method can be further promoted to cloud platform for lithium-ion batteries, fuel cell batteries and other energy storage system.