Automated Battery Making Fault Classification Using Over-Sampled Image Data CNN Features

Abstract

Due to the tremendous expectations placed on batteries to produce a reliable and secure product, fault detection has become a critical part of the manufacturing process. Manually, it takes much labor and effort to test each battery individually for manufacturing faults including burning, welding that is too high, missing welds, shifting, welding holes, and so forth. Additionally, manual battery fault detection takes too much time and is extremely expensive. We solved this issue by using image processing and machine learning techniques to automatically detect faults in the battery manufacturing process. Our approach will reduce the need for human intervention, save time, and be easy to implement. A CMOS camera was used to collect a large number of images belonging to eight common battery manufacturing faults. The welding area of the batteries' positive and negative terminals was captured from different distances, between 40 and 50 cm. Before deploying the learning models, first, we used the CNN for feature extraction from the image data. To over-sample the dataset, we used the Synthetic Minority Over-sampling Technique (SMOTE) since the dataset was highly imbalanced, resulting in over-fitting of the learning model. Several machine learning and deep learning models were deployed on the CNN-extracted features and over-sampled data. Random forest achieved a significant 84% accuracy with our proposed approach. Additionally, we applied K-fold cross-validation with the proposed approach to validate the significance of the approach, and the logistic regression achieved an 81.897% mean accuracy score and a +/- 0.0255 standard deviation.

Keywords: SMOTE; deep learning; fault detection; image classification; machine learning.

Figure 1

Proposed workflow diagram for battery fault detection.

Figure 2

Sample of battery fault images: (a) the right side shows the normal image and the left side shows the burn image; (b) the right side shows the cover is the wrong image, and the left side shows the image of the continuous hole; (c) the right side shows the weld too high image, and the left side shows the welding hole image; (d) right side shows the shifting image, and the left side shows the lack of weld image.

Figure 3

CNN architecture for feature extraction.

Figure 4

X-axis shows the Actual class and Y-axis shows the Predicted class where (a) represents the confusion matrix for Random forest-classifier, (b) represents the confusion matrix for Logistic Regression-classifier, (c) represents the confusion matrix for Support Vector Machine-classifier, (d) represents the confusion matrix for K-Nearest Neighbor-classifier.

Figure 5

Performance of ML classifiers: (a) represents the precision score, (b) represents the recall score, and (c) represents the F1 score.

Figure 6

Training and validation accuracy of the (a) CNN model using the imbalanced dataset, (b) CNN model using the balanced dataset, (c) DenseNet-121 model using the imbalanced dataset, (d) DenseNet-121 model using the balanced dataset, (e) MobileNet-V2 model using the imbalanced dataset, and (f) MobileNet-V2 model using the balanced dataset.

Figure 7

Training and validation loss of (a) CNN model using the imbalanced dataset, (b) CNN model using the balanced dataset, (c) DenseNet-121 model using the imbalanced dataset, (d) DenseNet-121 model using the balanced dataset, (e) MobileNet-V2 model using the imbalanced dataset, and (f) MobileNet-V2 model using the balanced dataset.