Multistage Manufacturing Processes Research Articles

Stream of variation modeling and monitoring for heterogeneous profiles in multi-stage manufacturing processes

ABTRACT Advanced manufacturing systems typically involve multiple operating stages, where in-process data are observed in the form of profiles that contain rich information for process monitoring. Emerging methodologies have been developed for multi-stage profile monitoring. However, the cases where different stages’ profiles are heterogeneous and variations propagate over stages have not been addressed so far. To this end, this article proposes a heterogeneous profiles-based stream of variation (HP-SoV) framework for multi-stage manufacturing process monitoring. It uses functional decomposition to extract features of heterogeneous profiles from different stages, where the decomposition coefficients are regarded as latent states that propagate along consecutive stages to capture the variation propagation. HP-SoV includes many existing models as its special cases, and enjoys an efficient inference algorithm via maximum likelihood estimation. Based on the one-step-ahead forecast errors of HP-SoV, a group monitoring scheme is further developed for online monitoring. Extensive numerical studies explore the modeling robustness and monitoring effectiveness of HP-SoV under different settings. A case study demonstrates the applicability of HP-SoV in multi-stage process monitoring for heterogeneous profiles.

A Partial Domain Generalization Method for Modeling Multiple Multistage Manufacturing Processes

In complex manufacturing systems, materials are processed by machines sequentially before final products are get, forming a multistage manufacturing process (MMP). In a modern massive production factory, it is common that multiple MMPs work simultaneously to fulfill productivity needs. Multiple MMPs may contain machines with different running time, processing the same types of products but with different specifications. In such a Multiple MMP(MMMP) system, it is critical for machines to learn from each other to gain optimal parameter settings of each. However, traditional machine learning methods usually fail to consider both similarities and unique features among multiple processes at the same time. To address this problem, a partial domain generalization method is proposed based on the thought of transfer learning to combine useful information from historical processes while maintaining features from the in-production process. The proposed method also suits for problems with the assumption that only partial input variables are available. Studies reveal that the proposed method has superior prediction performance over traditional machine learning methods and some widely-used transfer learning methods.

On the Potential of Manufacturing Multi-Material Components With Micro/Nanocellular Structures Via the Hybrid Process of Electromagnetic Forming Injection Foaming

Abstract Multimaterial design with a combination of solid and foam structures offers a promising avenue for reducing component weight while enhancing their functionalities. However, the complexity of multistage manufacturing processes poses significant challenges to adopting such approaches. To address these challenges, this paper introduces an innovative concept known as Electromagnetic Forming Injection Foaming (EFIF), which integrates injection molding, forming, and foaming processes into a single hybrid process. This process begins with a simultaneous filling-forming phase, followed by supercritical fluid (SCF) assisted foaming controlled by electromagnetic forming. Through a series of experimental and analytical studies, this work investigates the feasibility and effectiveness of EFIF. First, the impact of pressure drop rate and pressure drop on cell size and density is examined through a specialized experimental setup enabling performing injection, forming, and foaming processes in a single operation. The potential influence of electromagnetic forming on foam injection molding is explored through experiments focusing on the effects of a polymer layer between sheet metal blank and the electromagnetic coils. Additionally, an analytical study evaluates the EFIF process by calculating expected pressure drop rates under different processing conditions and their influence on cell nucleation rates. The results showed the possibility of achieving pressure drop rates up to 1.5 × 105 bar/sec, resulting in nucleation rates up to 1.77 × 109 nuclei/cm3sec. Overall, this paper highlights the potential of EFIF to merge existing technologies into a scalable solution for manufacturing multimaterial components with micro- to nanocellular polymer foams.

Tensor-based statistical learning methods for diagnosing product quality defects in multistage manufacturing processes

Multistage manufacturing processes with identical stages provide three-dimensional process data in which the first dimension represents the process (control/sensing) variable, the second is the stage, and the third is the measurement/sampling/data acquisition time point. Diagnosing quality faults in such processes often requires the simultaneous identification of crucial process variables and stages associated with product quality anomalies. Most existing diagnosis methods convert 3D data into a 2D matrix, resulting in loss of information and reduced diagnostic accuracy and stability. To address this challenge, we propose a penalized tensor regression model that regresses the product quality index against its 3D process data. For the estimation of high-dimensional regression coefficients with the limited amount of historical data, we apply the CANDECOMP/PARAFAC and Tucker decompositions to the coefficient tensor, which significantly reduces the number of parameters to be estimated. Based on the decompositions, a new regularization term is designed to enable the joint identification of critical process variables and stages. To estimate the parameters, we develop the block coordinate proximal descent algorithm and provide its convergence guarantee. Numerical studies demonstrate that the proposed methods can enhance diagnostic stability and on average improve the diagnostic accuracy by around 20% over existing benchmarks.

Robust Bayesian target vector optimization for multi-stage manufacturing processes

This research focuses on optimizing multi-stage manufacturing processes using Bayesian optimization (BO) with a robust Expected Improvement (EI) acquisition function. The aim is to optimize towards pre-selected target vectors, not to just minimize or maximize a function. To achieve this, we minimize the Euclidean distance between the actual and target output vectors, which requires transforming the Gaussian surrogate model posterior distribution into a non-central χ2(NCχ2) distribution. Furthermore, the distance measure additionally uses aleatoric uncertainty estimates of the actual output vectors to achieve robustness. We use a cascaded method that also considers the optimization results of intermediate stages, whereby optimization results are propagated from the last stage towards the first stage in each optimization iteration. By considering intermediate process outputs and aleatoric effects, our approach provides a robust optimization method for multi-stage manufacturing processes. To validate our method and to evaluate its properties, we use two artificial use cases. Moreover, we evaluate our approach in an industrial multi-stage forging process for the manufacturing of a nickel basis superalloy turbine disk, where involved stages are represented by corresponding 2D finite element method (FEM) DEFORM simulations. Evaluation suggests that our approach is superior in optimizing multi-stage manufacturing processes by considering all stage outcomes, robust distance measures, and the use of appropriate uncertainty distributions.

Multi-stage manufacturing process parameter optimization method based on improved marine predator algorithm

Product quality is a critical factor in manufacturing industry competition, and mechanical processing technology has been widely applied in manufacturing, directly affecting product quality. Therefore, it is very important to find the appropriate optimal parameters to improve the impact of processing on product quality. However, modern production processes are characterized by complex mechanisms and the mutual influence of multiple processes, which poses higher challenges for optimizing processing technology parameters. In this regard, the thesis proposes a method for optimizing process parameters in multi-process manufacturing based on an improved marine predator algorithm, aiming to optimize and improve process parameters in multi-process manufacturing processes. Firstly, a multi-process modeling strategy is adopted to explore the nonlinear relationship between process parameters and quality indicators based on multi-gene genetic planning, establishing a multi-process parameter optimization objective model. This effectively solves the problem of modeling difficulty caused by severe coupling of multiple processes. Then, to improve the efficiency of solving the optimization objective model, an improved marine predator algorithm is proposed, utilizing reverse learning strategies and mixed control parameters to enhance optimization capability, thereby obtaining the global optimal solution. Finally, using production process data from a certain factory as an example, the feasibility of the proposed method is verified, achieving the goal of multi-process process parameter optimization and ensuring the stability of product quality.

A Quality Prediction and Parameter Optimization Approach for Turbine Blade Multistage Manufacturing

As core components of an aircraft engine, turbine blades are featured by complex curved surface with strict precision requirements in multistage manufacturing processes (MMPs). Hence, reliable quality prediction and parameter optimization strategy on MMPs have long been in great demand. However, most techniques focus on the precision improving of single-stage single-product manufacturing systems; these improvements are difficult to implement in MMPs with coordinate transformations among multiple machines. In this article, we propose a noninvasive feature-based quality prediction and parameter optimization framework for high-precision MMPs. In particular, rather than retrofitting sensors onto machinery after installation, this study presents an executable and noninvasive solution by using coordinate measuring machine (CMM) data as input features for prediction. Based on features of different stages of machining, a sequential multistep deep learning architecture is developed to predict the final product quality. A preheating optimization algorithm, modified from a local gradient search method, is used for the optimization of key initial machining features, which results are further improved by the L-BFGS method, thus increasing the final production quality. To estimate the effectiveness of the proposed framework, experiments are carried out on two series of aviation turbine blades with complex curved surfaces from the Wuxi Turbine Blade Company Ltd. China. CMM data at key stages (i.e., blade root milling and comprehensive accurate milling) are used as inputs to predict the final geometrical errors. Comparison results verify that the proposed framework achieves the smallest root mean square error at 0.030 mm and 0.032 mm for two blades, respectively.

Multistage Net: Learning Continuous Multistage Manufacturing Processes of Liquid Products Without Intermediate Output and Lead-Time Labels

Multistage Net: Learning Continuous Multistage Manufacturing Processes of Liquid Products Without Intermediate Output and Lead-Time Labels

Sequential Modeling and Knowledge Source Integration for Identifying the Structure of a Bayesian Network for Multistage Process Monitoring and Diagnosis

Abstract As a popular applied artificial intelligence tool, Bayesian networks are increasingly being used to model multistage manufacturing processes for fault diagnosis purposes. However, the major issue limiting the practical adoption of Bayesian networks is the difficulty of learning the network structure for large multistage processes. Traditionally, Bayesian network structures are learned either with the help of domain experts or by utilizing data-driven structure learning algorithms through trial and error. Both approaches have their limitations. On the one hand, the expert-driven approach is costly, time-consuming, cumbersome for large networks, and susceptible to errors in assessing probabilities; on the other hand, data-driven approaches suffer from noise, biases, and inadequacy of training data and often fail to capture the physical causal structure of the data. Therefore, in this article, we propose a Bayesian network structure learning approach where popular manufacturing knowledge sources like the failure mode and effect analysis (FMEA) and hierarchical variable ordering are used as structural priors to guide the data-driven structure learning process. In addition, to introduce modularity and flexibility into the learning process, we present a sequential modeling approach for structure learning so that large multistage networks can be learned stage by stage progressively. Furthermore, through simulation studies, we compare and analyze the performance of the knowledge source–based structurally biased networks in the context of multistage process fault diagnosis.

MMP Net: A feedforward neural network model with sequential inputs for representing continuous multistage manufacturing processes without intermediate outputs

Machine learning models that are used for the prediction and control of production can improve quality and yield. However, developing models that are highly accurate and reflective of real-world processes is challenging. We propose a feedforward neural network model specifically designed for continuous Multistage Manufacturing Processes (MMPs) without intermediate outputs. This model, which is termed “MMP Net,” can accurately represent the control mechanism of continuous MMPs. Whereas existing studies on learning MMPs assume an intermediate output data, the MMP Net does not require such an unrealistic assumption. We use the MMP Net to develop prediction models for the lubricant base oil production process of a world-leading lubricant manufacturer. Evaluation results show that the MMP Net is superior to other deep neural network and machine learning models. Consequently, the MMP Net was actually implemented in a real factory in 2022 and is expected to save 900,000 dollars per year for each production line. We believe that our work can serve as a basis to develop customized machine learning solutions for improving continuous MMPs.

Development of combined hardening model for spring-back simulation of DP600 in multi-stage sheet metal forming

High strength steel DP600 is widely used in automobile industry due to its excellent mechanical property compared to conventional steel, but spring-back issue is considered to be the major challenge of DP600 in forming process, as the spring-back is triggered by the release of complicated internal stress. In multi-stage sheet metal forming, the sheet may be subject to several cycles of loading and reverse loading, Bauschinger effect greatly influence the strain stress behaviour. A well designed material model can be used to accurately predict and compensate spring-back in forming process. A combined hardening model especially designed for multi-stage forming is proposed in this paper, and the iterated compensation method could be applied by using LS-DYNA to get an optimized result. An experimental testing facility was developed for calibration of this model with cyclic tension and compression strain stress curve. The compensation method is applied in the numerical simulation of a multi-stage manufacturing process of an A-pillar, the result demonstrates the feasibility of the combined hardening model.

Make-or-buy strategies in a multi-stage manufacturing process and the role of learning effect in relocation decisions

Current global problems like the pandemic or the war in Ukraine have raised new aspects of make-or-buy decisions. Many companies consider shortening their supply chains or even reshore production. We are modelling the cost-related consequences of possible strategies, jointly considering the effect of productivity knowledge and interdependencies of consecutive production stages in the manufacturing process, assuming varying learning rates based on task complexity. The dynamic optimization problem enables us to investigate the role of learning effect in both outsourcing and insourcing (reshoring) decisions. In general, we find that firms should stick to or insource production processes where learning potential is high; however, keeping activities with low learning potential but higher interdependencies may also be beneficial as they can contribute to increase productivity at other stages. Optimal decisions might dynamically change from one period to the other. Reasons include that it might be reasonable to keep a production stage in-house until its cross-stage learning effect sufficiently contributes to decrease production costs at other stages. Accordingly, if a firm partly re-establishes its in-house manufacturing, the rest of the production stages may also be brought back over time thanks to the accumulated productivity knowledge. This may provide further implications for industrial policies that foster reshoring: pursuing activities with considerable learning potential or possible interdependencies with other corporate functions would trigger further relocation decisions and may result in upgrading in the value chain. We show numerical examples to illustrate the rationale behind different strategies, highlighting the key role of cross-stage learning.

A gray correlation based Bayesian network model for fault source diagnosis of multistage process – Small sample manufacturing system

Fault source diagnosis methodology is one of the key technologies of quality control and assurance for multi-source & multi-stage manufacturing processes, especially in small sample manufacturing systems. By analyzing the existing research on fault source diagnosis methods, a Bayesian network-based methodology is proposed. Gray correlation theory and mechanism analysis method are used in the process of Bayesian network model construction to reduce the dependence of sample data size for structure learning in the process of small sample manufacturing of complex products. In addition, two fault source diagnosis methods based on manufacturing principle analysis and reverse Bayesian network respectively are proposed. The strategy of the combined use of the two methods in the actual manufacturing scenes is given to cope with the fault source diagnosis scenario in the real manufacturing process. In the end, an example from an actual factory is provided to validate the effectiveness and efficiency of the proposed model and methodologies.

A methodology for data-driven adjustment of variation propagation models in multistage manufacturing processes

In the current paradigm of Zero Defect Manufacturing, it is essential to obtain mathematical models that express the propagation of manufacturing deviations along Multistage Manufacturing Processes (MMPs). Linear physical-based models such as the Stream of Variation (SoV) model are commonly used, but its accuracy may be limited when applied to MMPs with a large amount of stages, mainly because of the modeling errors at each stage that are accumulated downstream.In this paper we propose a methodology to calibrate the SoV model using data from the inspection stations and prior engineering-based knowledge. The data used for calibration does not contain information about the sources of variation, and they must be estimated as part of the model adjustment procedure. The proposed methodology consists of a recursive algorithm that minimizes the difference between the sample covariance of the measured Key Product Characteristic (KPC) deviations and its estimation, which is a function of a variation propagation matrix and the covariance of the deviation of the variation sources. To solve the problem with standard convex optimization tools, Schur complements and Taylor series linearizations are applied. The output of the algorithm is an adjusted model, which consists of a variation propagation matrix and an estimation of the aforementioned variation source covariance.In order to validate the performance of the algorithm, a simulated case study is analyzed. The results, based on Monte Carlo simulations, show that the estimation errors of the KPC deviation covariances are proportional to the measurement noise variance and inversely proportional to the number of processed parts that have been used to train the algorithm, similarly to other process estimators in the literature.

A novel predict-prevention quality control method of multi-stage manufacturing process towards zero defect manufacturing

A novel predict-prevention quality control method of multi-stage manufacturing process towards zero defect manufacturing

Optimizing the quality control of multivariate processes under an improved Mahalanobis–Taguchi system

Quality characteristics in manufacturing are correlated and do not follow a normal distribution. This study proposes a quality control method for multivariate manufacturing processes that are based on an improved Mahalanobis–Taguchi System (IMTS). The MTS has no data distribution assumptions and identifies anomalies through the Mahalanobis distance (MD). However, a covariance distance can consider the correlation between variables. Further, to address the shortcomings of the MTS in feature selection and threshold determination. A joint optimization model is proposed in this paper. Under this approach, the IMTS is employed to perform composite analyses on multiple quality characteristics and reduce dimensionality to identify abnormalities and the key quality characteristics that lead to anomalies. Further, various models are compared to construct the optimal non-parametric prediction models for each key quality characteristic. Finally, a conceptual model of process parameter optimization is proposed, which improves the Taguchi method to obtain the optimal combination of process parameters and their importance ranking, as the basis for process adjustment. By applying the proposed method, results show that the IMTS has an abnormality identification rate of 99.5%, which is higher than other methods such as MTS, support vector machine (SVM), back propagation neural network (BPNN), fast correlation-based filter solution SVM (FCBF-SVM) and sequential backward selection BPNN (SBS-BPNN). The dimensionality reduction rate is 0.5, which is higher than MTS, SVM, BPNN, and SBS-BPNN methods. The random forest (RF) algorithm is used for accurate predictions of all five key quality characteristics, the improved Taguchi method guided adjustments to manufacturing processes objectively, effectively, and economically.

A convex two-dimensional variable selection method for the root-cause diagnostics of product defects

Many multistage manufacturing processes consist of multiple identical stages. The root cause diagnostic of the product quality defects of these processes often involves the simultaneous identification of crucial stages and process variables that are related to product anomalies. In the literature, this is typically achieved by using penalized matrix regression that regresses the index of product defect against a matrix whose rows and columns respectively represent the stages and process variables. However, most existing models have some limitations that compromise their applicability and/or performance. For example, some models have an assumption on the rank of the coefficient, which often cannot be satisfied; some others formulate a nonconvex optimization criterion that easily results in a local optimum. Also, most models only provide diagnostics results with group-wise (i.e., stage- and variable-wise) sparsity. To address these challenges, this article proposes a novel convex two-dimensional variable selection method that can inspire both group-wise and element-wise sparsity. This is accomplished by proposing a new generalized matrix regression model and simultaneously penalizing the rows, columns, and elements of the regression coefficient matrix using an ℓ2, ℓ2, and ℓ1 norm, respectively. Simulated and real-world data are used to validate the effectiveness of the proposed method.

A novel machine learning-based multiobjective robust optimisation strategy for quality improvement of multivariate manufacturing processes

The primary objective of this study was to develop a novel data-driven machine learning-based multiobjective robust optimisation strategy to improve the overall quality of multivariate manufacturing processes. The new strategy was conceptualised considering a manufacturing environment with unreplicated non-normal data observations and limited opportunity for off-line sequential design of experiments. At a macro level, the new strategy adopts suitable artificial intelligence-based process models and a fine-tuned non-dominated sorting genetic algorithm-II (NSGA-II) to derive robust efficient process setting conditions. These robust solutions are iteratively derived considering process model predictive uncertainties, process setting sensitivities, and variance-covariance structure of uncontrollable multivariate non-normal inputs (or covariates). These solutions are also ranked based on multicriteria decision-making (MCDM) techniques to facilitate implementation. In this study, the quality of the best-ranked solutions was compared (w.r.t. closeness to specified multiple targets and predicted multivariate output variabilities) with those of the solutions obtained from parametric and commercial software-based approaches using three different real-life manufacturing cases.

Golden Path Search Algorithm for the KSA Scheme

The concepts of Industry 4.1 for achieving Zero-Defect (ZD) manufacturing were disclosed in <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">IEEE Robotics and Automation Letters</i> in January 2016. ZD of all the deliverables can be achieved by discarding the defective products via a real-time and online total inspection technology, such as Automatic Virtual Metrology (AVM). Further, the Key-variable Search Algorithm (KSA) of the Intelligent Yield Management (IYM) system developed by our research team can be utilized to find out the root causes of the defects for continuous improvement on those defective products. As such, nearly ZD of all products may be achieved. However, in a multistage manufacturing process (MMP) environment, a workpiece may randomly pass through one of the manufacturing devices with the same function in each stage. Different devices of the same type perform differently in each stage, where the performances will be accumulated through the designated manufacturing process and affect the final yield. KSA can only identify the influence of univariate variables (i.e., single devices) on the yield, yet it cannot detect the manufacturing paths that have significant influence on the yield. In order to cope with this deficiency such that the golden path with a better yield amongst all the MMP paths can be found, this research proposes the Golden Path Search Algorithm (GPSA), which can plan golden paths with high yield under the condition of the number of variables being much larger than that of samples. As a result, it makes the improvement of manufacturing yield be more comprehensive. <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Note to Practitioners</i> —Traditional scheduling only considers the capacity of the manufacturing devices for allocation; while, the impact on the yield is rarely considered. In fact, in a production process, the production deviations will be gradually accumulated and affect the product quality along with the processing influence of each device. Therefore, the purpose of this paper is to propose the GPSA scheme to quickly search for high-yield manufacturing paths before the production. Manufacturers can then configure production devices based on these paths. According to the experimental results of real manufacturers’ data, GPSA can not only quickly nail down the high-yield paths from a large amount of historical data, but also alert the users to avoid paths that are prone to defective rates for their reference.

A Product-Oriented Synchronization and Effective Information Extraction of Continuous Streaming Data for Relationship Mining in a Hot Rolling Process

Abstract Massive continuous streaming data are generated over time during production in a multistage manufacturing process. This paper aims to develop a product-oriented synchronization and effective information extraction of continuous streaming data and further model the relationships among variables for knowledge discovery. Take the steel rolling process as an example; this paper proposes a three-step data analytics procedure for product-oriented synchronization of continuous streaming data, effective information extraction, and further conducting relationship mining between the roll gap adjustment operations and product shapes based on the product-oriented data. The developed procedure first converts the continuous streaming data generated over time in a production process to product-oriented data set, then extracts the information related to the causes and effects of roll gap adjustments, and finally fits the model describing the relationship among the roll gap adjustments, the change of rolling torques, and the change of product dimensions. This data analytics procedure facilitates the decision-making in the steel rolling process and illustrates an effective application of massive in-situ sensing data towards intelligent decision-making in data-rich manufacturing processes.