Part Quality Prediction and Variation Reduction in Multistage Machining Processes Based on Skin Model Shapes
Time: Fri 2020-12-18 10.00
Location: https://kth-se.zoom.us/j/66765860566, Stockholm (English)
Subject area: Production Engineering
Doctoral student: Filmon Yacob , Tillverkning och mätsystem
Opponent: Univ.-Prof. Dr.-Ing. habil. Sophie Gröger, Technical University of Chemnitz
Supervisor: Assoc. Prof. Daniel Semere, Tillverkning och mätsystem
All manufacturing processes inevitably induce variations into manufactured parts that may result in nonconformance. Nonconforming parts incur costs due to the additional process required for rework or scrap loss. Hence, methodical efforts to reduce these variations are necessary for competitive manufacturing. To achieve this, effective variation reduction strategies have to be in place. In a multistage machining context, this could mean robust, rapid, and accurate approaches for representation and prediction of variations, change detection, variations source identification, and compensation.
Moreover, the approaches used should be capable of handling all forms of errors contributing to the propagation of variations and nonconformance. Existing part variation and variation propagation analysis methods for multistage machining are limited to orientation and position errors, neglecting form errors. Form errors can be captured by utilizing the concept of Skin Models Shapes (SMSs). The application of SMSs for multistage machining and variation reduction strategies has been limited and not established yet. This thesis contributes to developing and demonstrating the use of SMSs for part quality prediction and variation reduction in multistage machining processes.
The specific contribution of the thesis can be summarized as (i) the derivations of variation propagation models using dual quaternions; (ii) part quality prediction considering fixtures with locating surfaces, 3-2-1, and N-2-1 (N>3) locators; (iii) Octrees based method for performing statistical shape analysis; (iv) change and anomaly detection using machine learning classifiers; (v) variation source identification using pattern matching technique; (vi) and estimation of variation compensation values using dual quaternions.