Journal
IEEE SENSORS LETTERS
Volume 7, Issue 1, Pages -Publisher
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/LSENS.2022.3228327
Keywords
Sensor applications; explainable artificial intelligence (XAI); 3-D printer; fault diagnosis; ensemble learning; random forest; XGBoost
Ask authors/readers for more resources
This study proposes a data-driven approach for diagnosing faults in 3D printers. By collecting data for three scenarios - healthy condition, bed failure, and arm failure - an ensemble learning model of Random Forest and XGBoost achieves an accuracy of 99.75%. The interpretation of the model is improved using the Shapley additive explanations library.
Additive manufacturing is one of the most widely used techniques in the domain of manufacturing. Three-dimensional (3-D) printers are one of those systems that made additive manufacturing easier. Fused-deposition-modeling-based 3-D printers provide cost-effective 3-D models. Like other mechanical systems, 3-D printers also face faults that damage the printing of the system. Hence, proper maintenance is required. The data-driven-based approach of the diagnosis of the fault in 3-D printers is proposed in this letter. Data are collected for three scenarios-1) healthy condition, 2) bed failure, and 3) arm failure. The ensemble learning model of Random Forest and XGBoost has been implemented with a ratio of 0.54 and 0.46, and a result of 99.75% accuracy is achieved, compared to 96% and 98% alone, respectively. The black box machine learning model is then further explained using the Shapley additive explanations library for the interpretation of the prediction, such that the trustworthiness of the artificial intelligence model gets improved.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available