The capability of measuring melt pool variation is the key evaluating metal additive manufacturing quality. To measure the variation, a metrology architecture with in-situ melt pool measurement and an estimation module is required. However, it is a challenge to effectively extract significant features from the huge data collected by the in-situ metrology for quality estimation requirement. The purpose of this paper is to propose an intelligent metrology architecture with an in-situ metrology (ISM) module and an enhanced automatic virtual metrology (AVM) system. The ISM module can extract the melt pool features with a coaxial camera and a pyrometer. On the other hand, the AVM system is improved with a feature selection method to solve the issue of limited samples in the component modeling quality. The examples with different metals are adopted to illustrate how the system works for estimating surface roughness and density of components, and, in the future, the system can even serve as the feedback signal for adaptive control of the process parameters by layering in an additive manufacturing (AM) system.

Quality control is the key for the widespread adoption of metal additive manufacturing (AM). However, online quality estimation is challenging because high-frequency stream data derived from in situ metrology have to be processed in a timely manner to figure out the complicated interactions among material, machine, and part. To tackle such issue, this article proposes an intelligent AM metrology (IAMM) architecture to decouple and evaluate quality variations caused by material properties, machine issues, and process parameters when building an AM part. The IAMM architecture can also estimate the online layer-to-layer quality (e.g., roughness and density of an AM part) by applying the robust parameters derived from the uniform design (UD) method via the enhanced automatic virtual metrology technology in a parallel computing environment, as soon as the microfeatures of melt-pools and the macrofeatures of each layer are extracted. In addition, the associated indices and features in the IAMM architecture can also be used to evaluate the defects caused by machine issues in the given process parameters. The results of case studies validate the applicability of the IAMM architecture and show that the proposed estimation models are prospective for future closed-loop control in an AM process.

The due-date quotation is a key performance indicator for managing customer orders which would influence customer acceptance and/or the future potential lateness penalty. The production cycle time and allowance time are added and used as the due date of order. The objective is to maximize the hit rate which is the percentage of the orders fulfilled within the time limit of quoted due date. Under the framework of supervised machine learning, we explore the new developments in feature selection and the optimal decision tree to predict cycle time by using mixed-integer optimization. Cycle time allowance could be added to the predicted cycle time or incorporated in an optimization problem as a managerial decision variable. Case studies are used to demonstrate the effectiveness of this approach, and their performances are comparable to the other popular ensemble tree approaches, such as random forests and gradient boosting.

This paper presents position-based optimization methods to schedule production of automatic cells of a wheel manufacturing factory. Real-time schedule is challenging when a cell is interrupted by various order changes. Given a sequence of orders to be scheduled, it is sorted based on an earliest due day (EDD) policy, a mixed integer linear programming model is formulated, and then rolling-horizon optimization methods are used to timely find the near-optimal schedule by minimizing earliness and tardiness penalties with setup times of a manufacturing cell. In addition, an original schedule can be partial rescheduled with the preset order sequence by using the linear programming model. Experimental results show that the proposed method enables a wheel manufacturing cell to reschedule its 3 to 5 daily orders within cycle time of a rim when there exist order changes, e.g., rush orders and customized orders. Hence, these proposed methods are promising to promptly derive the near-optimal schedule for satisfying the objective of mass customization for industry 4.0.

The Vietnam Mekong River Delta (VMRD) is a climate-sensitive region characterized by diverse ecosystems, including extensive mangrove forests that protect against sea-level rise and contribute to global carbon sequestration. Accurate land cover classification in the VMRD is essential but remains challenging due to complex landscapes and dynamic environmental conditions. The primary objective of this study is to propose a semi-supervised deep learning framework that integrates satellite indices with multi-temporal remote sensing data to address key classification challenges, particularly in situations where ground truth data is limited, as compared to unsupervised and supervised machine learning methods. Our comparative analysis across different sample sizes (500 to 6000 ground-truth data points) reveals critical insights into model performance and scalability. Supervised models, including Random Forest (RF), Support Vector Machine (SVM), and Convolutional Neural Network (CNN), demonstrated strong performance when sufficient labeled data were available, with CNN achieving the highest accuracy (0.97 at 6000 samples). However, at minimal sample sizes (500 sample points), these supervised approaches exhibited substantial limitations, with accuracies dropping dramatically (RF: 0.75, SVM: 0.80, CNN: 0.81). Supervised models also showed overfitting tendencies compared to official land cover statistics. In contrast, the semi-supervised approach (SoC4SS-FGVC) achieves remarkably high performance at small sample sizes (0.92 accuracy with 500 sample points), demonstrating strength under minimal data availability. The framework also showed improved capability in distinguishing spectrally similar land-cover classes and detecting environmentally sensitive types such as mangrove forests. Cross-validation with official statistics confirmed the semi-supervised model’s superior effectiveness in delineating paddy rice fields and its resistance to overfitting. The performance analysis demonstrates that SoC4SS-FGVC provides a practical and cost-effective solution for land cover mapping, particularly in regions where extensive ground-truth data collection is prohibitively expensive or logistically challenging.

Motivated by recent advances in optimization-based machine learning algorithms, this research proposes a suboptimal explainable scheme (SES) for monitoring and reasoning machining outcomes by data-driven decision tree models. After machining, the time-frequency features in the wavelet package are extracted from the sensing data. The key features are then selected via a mixed-integer optimization method and served as inputs of the random-forests-based local search tree (RF-LST) or the optimal decision tree (ODT) for estimating machining outcomes. The machining tool wear and surface roughness estimation results indicate that this approach has interpretable features and structure. Under this scheme, the different aging stages of the manufacturing process could be visualized and utilized by using key features and their corresponding affine prediction functions.