The integration of Intelligent Welding Systems (IWS) in smart manufacturing leverages advancements in sensors, robotics, and artificial intelligence to optimize welding processes. However, in industry practice, we still face challenges such as sufficient data is not available for every manufacturing task, the costs associated with welding data annotation quality, and the risk of knowledge forgetting during the continual welding process. To tackle these issues, we developed an Adaptive Domain-Enhanced Transfer Learning (ADETL) framework that integrates self-supervised and continual learning strategies. This framework is adept at using incremental and unlabeled data for pre-training, in which we analyze the parameter space, loss landscape, and make the model understand the behaviour of knowledge transfer from diverse source domains. The ADETL framework improves the performance of defect classification, offering a promising solution to the challenges inherent in automatic, continuous welding operations.

Geometric and dimensional variations in objects are caused by inevitable uncertainties in manufacturing processes and often lead to product quality challenges. Failing to model the effect of object shape errors, i.e., geometric and dimensional errors of parts, early during the design phase inhibits the ability to predict such quality challenges. This consequently leads to expensive design changes after freezing of design. State-of-art methodologies for modelling and simulating object shape error have limited defect fidelity, data versatility, and designer centricity that prevent their effective application during the early design phase. To overcome these limitations, this paper presents a novel Morphing Gaussian Random Field (MGRF) methodology for object shape error modelling and simulation. The MGRF methodology models the spatial correlation in the deviations of the part from its nominal design using Gaussian Random Fields and then, utilises the modelled spatial correlations to generate non-ideal parts by conditional simulations. The MGRF methodology has (i) high defect fidelity enabling it to simulate various part defects including local and global deformations, and technological patterns; (ii) high data versatility allowing it to simulate non-ideal parts under the constraint of limited data availability and to utilise historical non-ideal part data of similar parts; (iii) designer centric capabilities such as performing ‘what if?’ analysis of defects of practical importance; and; (iv) the ability to generate non-ideal parts conforming to statistical form tolerance specification without additional modelling effort. The aforementioned characteristics enable the MGRF methodology to accurately model and simulate the effect of object shape variations on product quality during the early design phase. Practical applications of the developed MGRF methodology and its advantages are demonstrated using sport-utility-vehicle door parts and compared against state-of-art methodologies.

The paper proposes a novel approach, Object Shape Error Correction (OSEC), to determine corrective action in order to mitigate root cause(s) (RCs) of dimensional and geometric product shape errors. It leverages Deep Deterministic Policy Gradient (DDPG) algorithm to learn optimal process parameters update policies based on high dimensional state estimates of multi-station assembly systems (MAS). These policies can be interpreted in engineering terms as sequential corrective adjustments of process parameters that are necessary to mitigate RCs of product shape errors. The approach has the capability to estimate adjustments of process parameters related to fixturing and joining while simultaneously accounting for (i) RC uncertainty estimation, (ii) Key Performance Indicator (KPI) improvement, (iii) MAS design architecture; and, (iv) MAS inherent stochasticity. In addition, the OSEC methodology leverages a reward function parameterized by user interpretable functional coefficients for optimal tradeoff involving various corrections requirements. Benchmarking using an industrial, automotive cross-member assembly system demonstrates a 40% increase in the effectiveness of corrective actions when compared to current approaches.

Compliant, nonrigid parts are widely used in many industries today. Existing variation simulation analysis on the manufacturing part focuses on orientation and position deviation with part shape errors being largely omitted. This is a valid approach for the rigid part, but it is unrealistic and can be problematic for the compliant part. In this study, a new methodology has been introduced to compliant assembly early phase design to generate various probable variated manufacturing parts that conform to predefined tolerance specification or meet the certain industrial requirement. The proposed method is based upon the novel idea of a parametric space envelope, a purpose-designed variation tool constructed from parametric curves. Variation of embedded manufacturing part is linked to and controlled by a compact set of envelope’s boundary control points. Part variation instances are generated by simulating the control points’ movement in a systematic and efficient way. Importantly, simulated variations can be visualized in 3-D virtual space to provide user insight into part variation. The proposed methodology can help identify assembly high-risk regions, select proper fixtures, guide assembly engineering changes, and optimize assembly operations. An industrial case study on a deformable vehicle door hinge plate is presented to illustrate the methodology.Note to Practitioners—This article is motivated by two acute problems encountered in geometric variation simulation analysis for compliant parts in the early design stage. First, form errors of the nonrigid part are not fully captured in existing methods. Second, variation visualization can significantly enhance understanding of the geometric variation effects, but it is difficult to achieve under existing approaches. Inspired by the idea of a parametric space envelope, this article proposes a new methodology by building a variation tool to aid the task. Under the proposal, the geometric variation of the compliant part is indirectly modeled through the constructed variation tool. This indirect modeling enables capturing intrapart interactions that are the major source of the inaccuracy of existing methods. The simulated geometric variation can be visualized through the variation tool. In addition, the developed method does not rely on historical manufacturing experience. This can be valuable for early stage designer when such production information is not available or expensive to get. Furthermore, the proposed method can be integrated into existing computer-aided design (CAD), computer-aided manufacturing (CAM), and computer-aided engineering (CAE) systems to improve overall design quality and reduce reliance on multiple physical prototyping.

This article proposes a novel object shape error response (OSER) approach to estimate the dimensional and geometric variation of assembled products and then, relate these to process parameters, which can be interpreted as root causes (RC) of the object shape defects. The OSER approach leverages Bayesian 3-D convolutional neural networks integrated with computer-aided engineering simulations for RC isolation. Compared with the existing methods, the proposed approach: 1) addresses a novel problem of applying deep learning for object shape error identification instead of object detection; 2) overcomes fundamental performance limitations of current linear approaches for RC analysis (RCA) of assembly systems that cannot be used on point cloud data; and 3) provides capabilities for unsolved challenges such as ill-conditioning, fault-multiplicity, RC prediction with uncertainty quantification, and learning at design phase when no measurement data are available. Comprehensive benchmarking with existing machine learning models demonstrates superior performance with R2= 0.98 and MAE = 0.05 mm, thus improving RCA capabilities by 29%.

The paper proposes a novel Object Shape Error Response (OSER) approach to estimate the dimensional and geometric variation of assembled products and then, relate, these to assembly process parameters, which can be interpreted as root causes (RC) of the object shape defects caused by the assembly system. The OSER approach leverages Bayesian 3D-Convolutional Neural Networks for RC isolation. Compared with the existing methods, the proposed approach (i) addresses a novel problem of applying deep learning for object shape error identification instead of nominal shape identification (object detection); and, (ii) overcome significant limitations in current approaches for Root Cause Analysis (RCA) of assembly systems which are frequently ill-conditioned (collinear fault patterns) with high fault-multiplicity (frequently encountered in 6-sigma strategy) while quantifying the prediction uncertainty. The verification of the proposed methodology is conducted based on the CAE model of a multi-stage assembly process for automotive SUV door made of compliant parts. The results demonstrate performance with R-Squared (\pmbR2)=0.98 and Mean Absolute Error (MAE)=0.05 mm, thus improving RCA capabilities by 29% as compared to state-of-the-art methods.

Tolerance is an essential part of design and manufacturing. It plays a key role in product quality and manufacturing costs. Understanding and controlling production variations on key geometric features can provide firms with a competitive edge. A model to link production variation to tolerance is highly desirable but difficult to build, especially for deformable parts with complex surfaces. Inspired by an innovative idea of volumetric space envelope (constructed from a base parametric curve), this paper proposes a novel spatial tolerance model. In this proposal, a volumetric envelope is superimposed onto the target manufacturing part, whose deformation and deviation (during manufacturing or assembly) are viewed as spatial variation, and this variation is modeled and linked to movements of envelope's control points. This unique model design bypasses direct modeling of complex intrapart interactions, which is nonlinear in general and a major source of inaccuracy and low efficiency of many existing methods. The adopted indirect modeling brings many benefits. It can handle complex shapes and surfaces, and is able to take into account form errors. Also, it is capable of modeling both global and local variations observed in many practical cases. The new method is illustrated and verified through an example on a deformable vehicle door hinge plate. The proposed model shows application potential in every major stage of production, and makes possible to build a coherent cross-production life-cycle tolerancing framework from the early stage design, to manufacturing, to postproduction quality inspection.