Power battery consistency fault prediction method

Abstract

The invention discloses a power battery consistency fault prediction method, which comprises the steps of: 1, partitioning vehicle data which triggers the consistency fault through a density-based clustering algorithm, and dividing the vehicle data into different intervals; 2, dividing each interval into different working conditions; 3, constructing a cosine similarity-based single battery consistency poor fault prediction model; and 4, performing real-time prediction. According to the power battery consistency fault prediction method, mass data of the big data terminal platform are analyzed and mined in real time, so that the mass data of the big data terminal platform are effectively utilized, different working conditions can be considered in a targeted manner, battery consistency prediction is carried out, the prediction accuracy is improved, and the false recognition rate is reduced.

Description

technical field [0001] The invention relates to the field of power batteries, in particular to a method for predicting consistent faults of power batteries. Background technique [0002] In recent years, as people pay more and more attention to environmental and energy issues, new energy vehicles, as a low-carbon vehicle, have attracted more and more attention and favor. As a vital part of electric vehicles, batteries directly affect the performance of all aspects of the entire electric vehicle, but at the same time, batteries are also the most prone to failure. The failure of the battery system is often caused by the performance of one or several battery cells. Therefore, the consistency of the power battery pack directly affects the service life, cruising range and vehicle safety of the electric vehicle. [0003] At present, the commonly used method for predicting the consistent failure of the power battery is that the battery management system (BMS) will collect and calc...

AI Technical Summary

Problems solved by technology

However, this prediction method has the following shortcomings. First, the false recognition rate is relatively high.

By studying the actual vehicle data, it will be found that there are still many moments when the pressure difference exceeds this threshold, but the system does not alarm

Secondly, different working conditions will affect the charging and discharging of the battery. This method does not take into account the various operating conditions of the vehicle, but uses a unified threshold to judge, which is not comprehensive

Again, BMS only collects battery pack data for a period of time, and cannot store a large amount of historical data. Nowadays, many sensors with corresponding functions are installed on electric vehicles to collect various states of vehicle operation and batteries, and then transmit them to each electric vehicle through the network For big data terminal platforms, there are currently few methods for consistent fault prediction through in-depth analysis and mining of these large amounts of historical data

Images