Anomaly detection in Skin Model Shapes using machine learning classifiers

Abstract

The concept of Skin Model Shapes has been proposed as a method to generate digital twins of manufactured parts and is a new paradigm in the design and manufacturing industry. Skin Model Shapes use discrete surface representation schemes, such as meshes and point clouds, to represent surfaces, which makes them enablers to perform an accurate tolerance analysis and surface inspection. However, online inspection of manufactured parts through use of Skin Model Shapes has not been extensively studied. Moreover, the existing geometric variation inspection techniques do not detect unfamiliar changes within tolerance, which could be the precursors to the onset of the manufacturing of out of tolerance part. To detect the unfamiliar changes, as anomalies, and categorize them as systematic and random variations, some unique surface characteristics can be extracted and studied. Random surface deviations exhibit narrow normal distributions, and systematic deviations, on the other hand, exhibit wide, skewed, and multimodal distributions. Using those surface characteristics as key traits, machine learning classifiers can be used to classify deviations into systematic and random variations. To illustrate the method, multiple samples from a truck component manufacturing line were scanned and the collected 3D point cloud data was used to extract features. A prediction score of 97-100% can be achieved by decision tree, k-nearest neighbor, support vector machines, and ensemble classifiers. The purposed approach is expected to extend the existing online inspection approaches and applications of Skin Model Shapes in quality control.

A Survey on AI-Driven Digital Twins in Industry 4.0: Smart Manufacturing and Advanced Robotics

“…Supervised learning algorithms refer to machine learning methods in which models are trained using labels. Typical supervised learning methods used in digital twin include supper vector machine (SVM) [ 195 , 201 ], decision trees [ 86 , 93 , 94 ], k-nearest neighbors [ 102 ], convolutional neural networks (CNN) [ 103 , 135 , 202 , 206 , 270 ] and recurrent neural networks (RNN) [ 202 ]. In practice, data labeling can be an expensive task.…”

“…At the quality control stage, classical supervised ML models, such as ANN, decision tree, and SVM, were expected to detect or predict potential deformations and surface deviations in production [ 101 , 102 ]. Deep learning (DL) computer vision models, including residual and convolutional neural networks were deployed to recognize eventual quality issues during the automatic production and machining features of parts [ 103 , 104 ], which could be further utilized to enhance the quality and efficiency of assembly processes [ 108 ] ( E -factor), or retraced to the production planning stage in order to support decision making on the basis of historical production knowledge [ 109 ], as a “smart expert” in a collaborative environment ( SG -factor).…”

Smart Manufacturing with Artificial Intelligence and Digital Twin: A Brief Review

“…Object recognition of a smart gripper [41] Anomaly detection of surface deviations of a truck component [38] Fringe projection profilometry for 3D…”

“…The accuracy of the labeling and the feature selection affect the outcome of the learning algorithm. The algorithms that are generally found in digital twins include support vector machines [37], decision trees [18], k-nearest neighbors [38], convolutional neural networks and recurrent neural networks [39].…”

Design and management of digital manufacturing and assembly systems in the Industry 4.0 era

“…Yacob et al [58] propose the concept of Skin Model Shapes as a method to generate digital twins of manufactured parts as a new paradigm in the design and manufacturing industry. Skin Model Shapes use discrete surface representation schemes to represent surfaces, accurate tolerance analysis and surface inspection.…”