Plug Valve Surface Defects Identification through Nondestructive Testing and Fuzzy Deep-Learning Algorithm for Metal Porosity and Surface Evaluation

Table 1

Applicability and novelty of the proposed work.

Technique	Applicability	Remarks
Machine learning [6]	Difficulty for real-time application	Nonavailability of real-time datasets for implementation. Learning process is slow, with issues in reusing, integrating, and debugging the models.
Neural network [7]	Not applicable	The training time is quiet long. There are chances for misclassification to occur.
Deep learning	Applicable	Automated feature extraction and classification. Training time is less with no misclassifications. Works effectively on real-time plug valve images.