The sources of metal foreign matter in the production line mainly include the following two aspects: one is the direct contact introduction of equipment and materials (direct introduction). The second is the introduction of metal scatter in the air into the material (indirect introduction).
This research results are helpful to improve the battery safety from the manufacturing side. Feature selection is conducted with feature importance analysis by using the RF method and OOB error calculation. Two features K2 and OCV4 which have the highest feature importance are selected from the 12 features and used as the input for FMD detection.
Detection of internal defects in lithium-ion batteries using lock-in thermography Blister defect detection based on convolutional neural network for polymer lithium-ion battery Process-product interdependencies in lamination of electrodes and separators for lithium-ion batteries
The first known application of the data-driven algorithms to solve the foreign matter defect detection problem. Experiments are conducted with implanted foreign matter defect on battery pilot manufacturing line. The proposed method achieves high precision, accuracy and recall compared with traditional HiPot test.
One of the core indicators of the battery cathode material production line. The sources of metal foreign matter in the production line mainly include the following two aspects: one is the direct contact introduction of equipment and materials (direct introduction).
Since the number of FMD batteries is far less than the number of the normal batteries, the weight of the two classes are adjusted to be inversely proportional to the class frequencies of the input data. The number of DT is set to 30 which is large enough to ensure OOB error calculation.