Early Warning of Thermal Runaway for Lithium-Ion Battery Based on Multi-Sensor Detection

The expansion of the electric vehicle market, giving credit to the technical breakthroughs in the energy density and cycle life of lithium-ion batteries, is beyond our expectations. However, the lithium-ion batteries with high energy density still suffers from safety problems, such as thermal runaway. Thermal runaway always occurs with fire and explosion, which threatens human lives and properties. It is essential to make early warning before the triggering of battery thermal runaway happens. In this study, an online fault-diagnosis algorithm that can be used for the early warning of thermal runaway is proposed based on joint information measured by multiple sensors. A set of sensors, including voltage, temperature, flammable gas and pressure, are used to monitor the direct and indirect features of thermal runaway. Figure 1 shows the scheme of the proposed fault-diagnosis algorithm, including five steps that explain how this algorithm makes decision of warning. Step I is for data collection, while the data are acquired from the thermal runaway tests. Step II is for data screening. The average temperature, average voltage, maximum temperature and minimum voltage are obtained for further data processing. Step III is for the difference calculation. The voltage difference and temperature difference are calculated. Step IV is called the fault level transformation. The continuous deviations and the continuous signals of gas sensor and pressure sensor are transformed into discretized fault levels. Step V is for the fault judgement. The overall fault level is calculated by merging all the fault levels. The thermal runaway can be captured by setting proper threshold to judge the overall fault level. To confirm that the algorithm is capable of detecting thermal runaway before it develops into a severe hazard, thermal runaway tests on commercial 18650 cells were conducted. A customized transparent PMMA box, which has comparable volume of the module of Tesla Model S, is designed as the experimental vessel. Battery cells, which have the identical cell number of one Tesla module, are placed in the vessel. Thermal runaway is triggered by heating one of the cells. Sensors are placed in different locations of the vessel to find the optimal location. The test results validate the warning algorithm, and will be presented at the conference. Acknowledgement: This research is supported by the National Natural Science Foundation of China under the Grant No. U1564205 and No.51706117, and funded by the Ministry of Science and Technology of China under the Grant No. 2016YFE0102200. Figure 1. The scheme of the proposed fault diagnosis algorithm for TR warning. Figure 1