Abstract
Internal short circuit that occurs inside lithium-ion batteries is known as one of the causes of thermal runaway. If micro-internal short circuits can be detected, the anomalies at very early stage can be known, and it will contribute to improved safety when using lithium-ion batteries. The purpose of this study is to fabricate a software architecture that can detect micro-internal short circuits of lithium-ion batteries during flight with a view to application to electric aircraft that require high safety. In this research, we first prepared a new lithium-ion battery and another lithium-ion battery of the same model that was intentionally deteriorated to make it easy to cause an internal short circuit. Next, we designed four features which denote a characteristic voltage behavior when the micro-internal short circuit occurs. A large feature value was obtained from the deteriorated battery, while such a value could not be obtained from the new battery. Therefore, we considered new batteries to be normal specimens, and tried to detect the abnormality of the deteriorated battery by k nearest neighborhood method. As a result, it was shown that micro-internal short circuits of lithium-ion batteries can be detected based on the features describing the behavior of the abnormal voltage change by the micro-internal short circuit.
