Measurement and 3D data acquisition methods
The traditional measuring methods apply measuring techniques that are operated by using special metrology devices such as inspection fixtures. These time-consuming techniques require skillful operators. However, developments in 3-D scanning technology allow creating a digital scan model from a physical object. Concerning the developed measuring systems (Savio, De Chiffre et al. 2007) and specifically laser scanners (Martínez, Cuesta et al. 2010), these measuring devices (scanners) can be categorized as contact and non-contact scanners. Contact scanners are based on Coordinate Measuring Machine (CMM) technology that can be controlled either manually or automatically by a program.
These devices consist of a probe that can move along three axes, where each axis has a built-in reference standard. The advantages of these scanners are: They are not sensitive to colors or transparency. These devices have high accuracy and low cost. They are capable of measuring deep slots and pockets.
Unfortunately, the data collection in these type of measuring devices is slow, and the probe contact can disturb the measurement of flexible parts by causing unwanted deformation during the measuring process (Leake and Borgerson 2013).
Non-contact scanners use lasers and optics (using charged-coupled device (CCD) sensors) to digitally capture the geometrical shape of a part as point clouds. The non-contact scanners are fast whereas there is no physical contact between the scanners and parts. Point clouds acquired by noncontact scanners have lower accuracy compared to measured data obtained by contact scanners. However, the accuracy of acquired point clouds via noncontact scanners for common inspection applications is quite acceptable. The transparency, reflectivity and in some cases the color of surfaces also can introduce limitations for noncontact scanners, which can be resolved by applying temporary non-reflective paints. These limitations can add noise in the acquired point clouds for which the robustness of relevant inspection methods needs to be validated.
Dimensioning, tolerancing and inspection specification of non-rigid parts
A dimension is a numerical value to define the size, location, geometrical characteristic, or surface texture of a part or features on a part. Based on ASME Y14.5, dimensions are applied to support the mating and functionality of parts. Mass-produced parts of commercial products are manufactured by applying the same production process for each part. Then, these manufactured parts are randomly assembled together. Therefore, these parts must be interchangeable to achieve a successful assembling process. However, due to the uncertainty of manufacturing processes two parts can never be the same. Therefore, slight variations in the part size are considered to prevent interference or loose assembling conditions. Tolerancing is a dimensioning technique to ensure part interchangeability by controlling the variation that exists in manufactured parts. The tolerances come off by specifying a range within which a dimension is allowed to vary. In other words, tolerancing ensures the functionality of parts and quality of production. Tolerancing can be expressed as the following (Leake and Borgerson 2013): A direct tolerancing method, which includes the limit dimensioning and plus-minus tolerancing. General tolerancing notes, to generally address a tolerancing for all dimensions. The Geometric Dimensioning and Tolerancing (GD&T), to verify the conformity of manufactured parts with the specification defined at the design stage.
GD&T are widely applied for manufactured parts with complex shapes in different industrial disciplines. Tolerance allocation and analysis methods are used in GD&T to assign proper tolerances for assembly processes. Unlike traditional methods, non-rigid parts tolerancing methods must take into consideration compliance and permissible displacements of non-rigid parts during the assembly process. The pioneer researches on tolerance analysis of non-rigid parts are carried out by (Liu, Hu et al. 1996, Liu and Hu 1997, Camelio, Hu et al. 2002) in which tolerance analysis of non-rigid parts for advanced vehicle manufacturing is established. (Merkley 1998, Mounaud, Thiebaut et al. 2011, Chen, Jin et al. 2014) present a review of studies on Computer-Aided Tolerancing (CAT), tolerance analysis and allocation strategy for compliant (non-rigid) parts. In this context, the profile tolerances are assigned to free-form surfaces of parts to control surface variations. These profile tolerances can be defined with reference to datum(s) known as related profile tolerances. Related profile tolerances are applied for cases that involve the assembly of free-form surfaces with other geometric features (Li and Gu 2005).
The primary step in a CAI method for non-rigid parts is a rigid registration. Rigid registration brings the CAD and scan models as close as possible in a common coordinate system without deforming the models. In fact, rigid registration uses an optimal transformation matrix to translate and rotate the models without making any changes to their shapes. Among different rigid registration methods such as those described by (Li and Gu 2004, Savio, De Chiffre et al. 2007), the Iterative Closest Point (ICP) algorithm (Besl and Mckay 1992) is widely applied in different domains and well-known as a robust and efficient rigid registration. For example, a rigid registration for inspection of aircraft parts is presented in (Ravishankar, Dutt et al. 2010) for which the modified ICP method is applied. Among all rigid registration methods, ICP algorithm is known as a statistically robust and efficient method. The optimal transformation (translation and rotation) matrix in ICP registration is estimated at each iteration to minimize the distance between two models. To this end, the Hausdorff distance (Henrikson 1999) between CAD mesh and acquired point clouds of scan model measures how far these two models are from each other.
Fixtureless inspection of non-rigid parts (non-rigid registration)
Applying only the rigid registration methods such as ICP method (Ravishankar, Dutt et al. 2010) is not sufficient for inspection of non-rigid partsin a free-state. Rigid registration methods do not permit any modification on the shape of CAD or scan models to compensate for the flexible deformation of non-rigid parts during the inspection process in a free-state. Meanwhile, inspection methods for non-rigid parts need to distinguish between the flexible deformation of non-rigid parts in a free-state with defects such as geometrical deviations due to faulty manufacturing. Conventional CAI methods apply restraining non-rigid parts in physical fixtures (Ascione and Polini 2010) during the inspection process. However, these sophisticated and expensive fixtures have significant drawbacks for which the setup and repeatability of the fixtures are costly. Therefore, the fixtureless inspection methods are required to eliminate these costly fixtures. Fixtureless inspection of non-rigid parts based on CAI methods is able to evaluate the geometrical deviation of manufactured parts with respect to the assigned tolerance by taking into consideration the flexible deformation of non-rigid parts in a free-state. A categorization of fixtureless inspection methods based on literature is presented in (Abenhaim,Desrochers et al. 2012). The fixtureless CAI methods are classified into four approaches as I) automated vision inspection, II) metric characteristic, III) boundary reconstruction and IV) simulated displacement. Fixtureless inspection of non-rigid parts can be performed by non-rigid registration methods classified as simulated displacement.
Verification and validation methods based on ASME recommendations
All CAI methods, which are based on scan data and computational calculations, are required to be verified and validated. Uncertainty in computational simulations and measuring errors due to the inaccuracy of data acquisition devices are inevitable in fixtureless non-rigid inspection methods. Inaccuracy of scanners is due to the technical limits of devices, optical effects (such as light fraction and reflectivity of parts surface) or inaccessible features of parts. These noisy data can affect the performance of CAI methods. Applying Verification And Validation (V&V) approaches on computational simulation models allows assessing accuracy, reliability, and robustness of simulations (Schwer, Mair et al. 2012). In fact, verification assesses the accuracy of a solution in a computational model whereas validation evaluates the consistency of computational simulation results comparing with the actual results. due to various sources of uncertainty in computer codes and simulations, all computational methods, among which are CAI methods, need to be thoroughly verified and validated. Investigations on the application and theory of verification and validation in computational engineering are presented by (Oberkampf, Trucano et al. 2004, Oberkampf and Barone 2006, Sornette, Davis et al. 2007). Unlike the qualitative traditional validation activities, newly developed rigorous and systematic V&V approaches are applied in different disciplines. These approaches are published for V&V guides in computational solid mechanics (Hills and Trucano 1999), fluid dynamics (Committee 1998), heat transfer (Committee 2009) and material engineering (Cowles, Backman et al. 2012). The predictive capability of the computational model in these guides is demonstrated by comparing the features of interest (validation metrics) with respect to the real model in its intended use. The validation metrics, which are the quantitative measure of agreement between a predictive model and physical observation (measurements), are categorized as hypothesis testing-based and distance-based (Oberkampf and Trucano 2008, Liu, Chen et al. 2011). Unlike distance-based metric, the hypothesis testing-based metrics provide an evaluation concerning the general trend of data. The hypothesis testing-based validation metrics compare the probability distributions of the prediction results and observations.
The comparison, which is based on statistical distributions such as cumulative distribution functions (CDFs), is accomplished through a test of significance enclosed in the hypothesis testing theory. Several research work, in various engineering fields such as structural dynamics (Paez and Urbina 2002), steady and transient heat conduction and shocks (Hills and Leslie 2003), thermal decomposition of polyurethane foam (Rutherford and Dowding 2003, Dowding, Leslie et al. 2004) and sheet metal forming processes (Chen, Baghdasaryan et al. 2004) used hypothesis testing-based validation metrics for their numerical calculations.
Table des matières
CHAPTER 1 INTRODUCTION
1.1 Structure of the thesis
CHAPTER 2 LITERATURE REVIEW
2.2 The compliance of non-rigid parts
2.3 Measurement and 3D data acquisition methods
2.4 Dimensioning, tolerancing and inspection specification of non-rigid parts
2.4.1 Rigid registration
2.4.2 Fixtureless inspection of non-rigid parts (non-rigid registration)
188.8.131.52 Fixtureless inspection based on virtually deforming the scan model
184.108.40.206 Fixtureless inspection based on virtually deforming the CAD model
2.5 Verification and validation methods based on ASME recommendations
2.6 State of the art summary
CHAPTER 3 GENERAL PLANNING OF THE THESIS
3.1 Statement of the problem
3.2 Research objectives
3.3 Hypotheses used in the project
3.4 The synthesis of researches and logical links between articles
3.4.1 Simulated scan models for validation cases
3.4.2 Automatic fixtureless CAI based on filtering corresponding sample points (Article 1)
3.4.3 Validation and verification of our CAI method (Article 2)
3.4.4 Virtual inspection in assembly-state using permissible loads for deviated non-rigid parts (Article 3)
CHAPTER 4 AUTOMATIC FIXTURELESS INSPECTION OF NON-RIGID PARTS BASED ON FILTERING REGISTRATION POINTS
4.3 Literature review
4.4 Methodology and implementation
4.4.1 Description of the proposed methodology
4.4.3 Validation on a case with no defects
4.5.1 Introduction: validation cases
4.5.2 Validation cases for part A
4.5.3 Validation cases for part B
CHAPTER 5 ASSESSMENT OF THE ROBUSTNESS OF A FIXTURELESS INSPECTION METHOD FOR NON-RIGID PARTS BASED ON A VERIFICATION AND VALIDATION APPROACH
5.3 Background on the approach to fixtureless CAI for non-rigid parts
5.4 Assessing the robustness of our CAI method based on ASME V&V recommendations
5.4.1 ASME recommendations for verification and validation
5.4.2 Verification and validation methodology for CAI
5.4.3 Robustness of our CAI method
5.5 Validation results for cases with small free-state deformation
5.5.1 Validation cases considered
5.5.2 Results for part A
5.5.3 Results for part B
5.5.4 Conclusions about validation cases for part B
5.6 Effect of large free-state deformation
5.6.1 Cases considered and results obtained
5.6.2 Conclusions about the effect of large free-state deformation
5.7 Discussion about results
CHAPTER 6 FIXTURELESS INSPECTION OF NON-RIGID PARTS BASED ON VIRTUAL MOUNTING IN AN ASSEMBLY-STATE USING PERMISSIBLE LOADS
6.3 Literature review
6.4 Virtual Mounting Assembly-State Inspection method (VMASI)
6.4.1 Pre-registration and partition of the scan mesh
6.4.2 Non-rigid registration using restraining pressures optimization (RPO)
6.4.3 Inspection and evaluation
6.4.4 The proposed VMASI algorithm
6.5 Application of proposed VMASI method on real parts
6.5.1 Introduction: validation cases
6.5.2 Results for part A
220.127.116.11 Scan models of part A with defects generated by geometric alteration
18.104.22.168 Scan models of part A with defects simulated by plastic deformation
6.5.3 Results for part B
CHAPTER 7 GENERAL DISCUSSION
7.1 Discussion on the sample points filtering method
7.2 Discussion on nonlinear FEA
7.3 Discussion on our developed V&V method
7.4 Discussion on our developed virtual mounting method
CHAPTER 8 CONTRIBUTIONS, PERSPECTIVES, AND CONCLUSIONS
8.1 Major contributions