Modeling the uncertainty from data is an essential quest in the learning of neural network models but has not been well addressed. A probabilistic neural network with Gaussian-mixture distributed ...parameters is developed in this work to provide an efficient and high-fidelity solution for learning multimodal uncertainties in neural networks. An adaptive Gaussian mixture scheme is adopted to refine the Gaussian mixture probability distributions and ensure the fidelity of uncertainty propagation in both linear and nonlinear transformations through the network. As its predictive distribution can be inferred analytically, this probabilistic network can be trained efficiently using a backpropagation method based on gradient descent. The proposed network not only achieves a state-of-the-art performance when benchmarked on a series of public datasets but also improves the accuracy and uncertainty quantification quality in two manufacturing process monitoring schemes. In a tool wear monitoring scheme for machining, it reduced the root mean square error (RMSE) by 44% and narrowed the confidence intervals of tool wear prediction by 35% compared to a neuro-fuzzy model. In a porosity monitoring system for additive manufacturing, the proposed network improved the porosity detection accuracy by 2% to 93.6% and quantified confidence intervals that were not available in conventional deep learning models. All these successes prove that the proposed probabilistic neural network can be a promising solution to address practical problems subject to significant uncertainties.
•General-form probability distributions of uncertainty are characterized with Gaussian mixtures.•Predictive distributions of the network can be inferred directly and analytically.•A sampling-free backpropagation method to learn probabilistic parameters is presented.•Remarkable improvement was achieved on two manufacturing process monitoring schemes.
Aroma is one of the most important criteria of tea quality, but the dynamic changes of aroma profile during the manufacturing process, and the chemical basis of characteristic aroma in Fu brick tea ...remain largely unknown. In this study, a total of 72 volatiles were identified and quantified, only the esters content increased sharply during the process. Sensory quantitative description analysis revealed that the ‘green’ attribute was dominated in the early processing stage, and the ‘fungal flower’, ‘flower’, ‘mint’ and ‘woody’ attributes became the major contributors to the aroma character in the later processing stages. Indicated by partial least-squares analysis, the linalool, acetophenone, and methyl salicylate were identified as key volatiles contributors to the ‘fungal flower’, ‘flower’, and ‘mint’ attributes, the cedrol contributed to ‘woody’ attribute, and twelve alcohols and aldehydes were related to ‘green’ attribute. Besides, bidirectional orthogonal partial least squares analysis revealed that six fungal genera Aspergillus, Candida, Debaryomyces, Penicillium, Unclassified_k_Fungi, Unclassified_o_Saccharomycetales were identified as core functional microorganisms link to the metabolism of volatiles. Taken together, these findings provide new insights into Fu brick tea aroma profile variation and increase our understanding of the formation mechanism of the characteristic aroma during the manufacturing process.
•Aroma and sensory profiles during Fu brick tea processing were clarified.•Key aroma compounds and microorganisms were identified.•Relationships among volatiles, sensory descriptors and microorganisms were exhibited.
Lyocell fiber has emerged as an important class of regenerated cellulose that is produced based on the N-methyl morpholine-N-oxide (NMMO) dissolution method, and it has unique properties compared to ...viscose fiber. The NMMO technology provides a simple, resource-conserving, and environmentally friendly method for producing regenerated cellulose fiber. In this paper, the manufacturing process, environmental impact, and product quality of lyocell fiber are reviewed and compared with those of the conventional viscose fiber.
Oxide-type all-solid-state lithium-ion batteries are considered as a promising candidate for next-generation batteries because they do not generate toxic gas like their sulfide-type counterparts. ...However, batteries based on oxide systems exhibit low performance because of several challenges, such as the low ionic conductivity of the solid electrolyte and poor contact between the electrode materials and the solid electrolyte. In this study, an all-inorganic oxide solid lithium-ion battery with very safe performance was manufactured by a room-temperature process and investigated. The 50Li2SO4–50Li2CO3 glass electrolyte was employed as a highly deformable electrolyte to improve the interparticle contact between the electrode active materials and electrolyte. An all-oxide solid-state LiNi1/3Co1/3Mn1/3O2–Li4Ti5O12 cell employing a 50Li2SO4–50Li2CO3 glass electrolyte was successfully prepared by cold pressing and achieved 135 mAh g−1 at 23 μA cm−2 and 60 °C. The calculated energy density of the cell was about 50 Wh kg−1 based on the masses of the positive and negative electrodes and the separate layer. Furthermore, despite its thinner separate layer than that of other cells, it exhibited stable cycle performance without a short circuit that was caused by lithium dendrite growth.
•Glass state LSCO showed excellent deformability and high ionic conductivity.•All-oxide-solid-state NCM-LTO battery using LSCO was prepared by only cold press.•The cell indicates discharge capacity of 135 mAh g−1 at 23 μA cm−2 and 60 °C.•It showed stable cycling performance despite the relatively thinner SE layer.
Optimization and control of Manufacturing Processes seem to be the key-enabling approaches towards Efficient Manufacturing and Zero-defect. Thus, it is of high importance to study what-if scenarios ...regarding changes of Process Parameters. To achieve this in real time, Digital Twins have to be designed and implemented. Towards this end, in the current work, a specific methodology, originating from process physics is reviewed as a candidate technology for the process level Digital Twin. The requirements are stated and checked one-to-one and respective numerical results are shown. Discussion on Zero-defect and optimization under the prism of Industry 4.0 is presented
In the era of the fourth industrial revolution, technological frameworks aim to arrange information and communication technologies into valuable assets to support the management of operations. One of ...these frameworks is the digital twin, which enables abstractions of a physical resource in the virtual space, allowing for behavior simulations and performance assessments. Consensus regarding which key features manufacturing digital twins should encompass has not yet been reached. This study summarizes a variety of features proposed by recent models in the literature. Four features are identified, followed by expert analyses that assess how valuable they are to the success of implementations.
Soft Robots Manufacturing: A Review Schmitt, François; Piccin, Olivier; Barbé, Laurent ...
Frontiers in robotics and AI,
2018, Volume:
5
Journal Article
Peer reviewed
Open access
The growing interest in soft robots comes from the new possibilities offered by these systems to cope with problems that cannot be addressed by robots built from rigid bodies. Many innovative ...solutions have been developed in recent years to design soft components and systems. They all demonstrate how soft robotics development is closely dependent on advanced manufacturing processes. This review aims at giving an insight on the current state of the art in soft robotics manufacturing. It first puts in light the elementary components that can be used to develop soft actuators, whether they use fluids, shape memory alloys, electro-active polymers or stimuli-responsive materials. Other types of elementary components, such as soft smart structures or soft-rigid hybrid systems, are then presented. The second part of this review deals with the manufacturing methods used to build complete soft structures. It includes molding, with possibly reinforcements and inclusions, additive manufacturing, thin-film manufacturing, shape deposition manufacturing, and bonding. The paper conclusions sums up the pros and cons of the presented techniques, and open to developing topics such as design methods for soft robotics and sensing technologies.
Life cycle assessment (LCA) plays a crucial role in green manufacturing to uncover the critical aspects for alleviating the environmental burdens due to manufacturing processes. However, the scarcity ...of life cycle inventory (LCI) data for the manufacturing processes is a considerable challenge. This paper proposes a novel approach to extrapolate LCI data of manufacturing processes. Taking advantage of LCI data in the Ecoinvent datasets, decision tree-based supervised machine learning models, namely decision tree, random forest, gradient boosting, and adaptive boosting, have been developed to extrapolate the data of GHG emissions, i.e., carbon dioxide, nitrous oxide, methane, and water vapor. Initially, a correlation analysis was conducted to derive the most influential factors on GHG quantities resulting from manufacturing activities. First, the collected data have been preprocessed and split into train and test sets (70% and 30%, respectively). Second, a five-fold cross-validation method was applied to tune the hyperparameters of the models. Then, the models were re-trained using the best hyperparameters and evaluated using the test set. The results reveal that the Gradient Boosting model has a superior predictive performance for extrapolating the GHG emission data, with average coefficients of determination (R2) on the test set <0.95. Moreover, the model predictions involve relatively low values of the average root mean squared error and an average mean percentage of error on the test set. The correlation and feature importance analyses emphasized that the workpiece material and manufacturing technology have a considerable effect on natural resource consumption, i.e., energy, material, and water inflows into the process. Meanwhile, energy consumption, water usage, and raw aluminum depletion were the most influential factors in GHG emissions. Eventually, a case study to extrapolate the inflows and the outflows for new manufacturing activities has been conducted using the validated models. The proposed GraBoost model provides a computational supplementary approach to estimate and extrapolate the GHG emissions for different manufacturing processes when LCI data are incomplete or don't exist within LCI databases.
•Extrapolation of LCI data using decision tree-based machine learning models.•Filling some gaps for manufacturing process LCI data.•Gradient Boosting model can bridge and extrapolate greenhouse gas emissions for new manufacturing activities.
•The most remarkable changes in volatiles of WRT occurred during roasting.•Flavanols were promptly oxidized and converted to heterogeneous oxidation products.•Flavonols, phenolic acids, and xanthine ...alkaloids remained stable during processing.•Central vacuole of tea leaf cells shrunk remarkably during withering.
Wuyi Rock tea (WRT), a top-ranking oolong tea, possesses characteristic woody, floral, nutty flavor. WRT flavor is mainly formed during the manufacturing process. However, details regarding its formation process are not fully understood yet. In this study, the dynamics of volatile and phenolic components over the whole manufacturing process of WRT were investigated. During withering, despite minor changes in volatile and phenolic components, the central vacuole shrunk remarkably, which reduced the cell mechanical performance and facilitated the subsequent enzymatic fermentation. During fermentation, approximately 78% of flavan-3-ols in fresh tea leaves were oxidized and converted to a diverse mixture of highly heterogeneous oxidation products, such as theaflavins, whereas flavonols, phenolic acids, and xanthine alkaloids remained stable throughout the manufacturing process. Aldehydes, ketones, and heterocyclic compounds, imparting woody, floral, and nutty scent, were mainly formed during the roasting steps. This detailed information can expand our understanding on the formation of WRT flavor.
Hybrid additive-subtractive manufacturing processes are becoming increasingly popular as a promising solution to overcome the current limitations of Additive Manufacturing (AM) technology and improve ...the dimensional accuracy and surface quality of parts. Surface roughness, as one of the most important surface quality measures, plays a key role in the fit of assemblies and thus needs to be thoroughly evaluated at the design and manufacturing stages. However, most of the studies on surface roughness modelling and analysis employ empirical approaches, and only consider the effect of a single manufacturing process. In particular, the existing surface roughness models are not applicable to hybrid additive-subtractive manufacturing processes in which a secondary process is involved. In this article, analytical models are established to predict the surface roughness of parts fabricated by AM as well as hybrid additive-subtractive manufacturing processes. A novel surface profile representation scheme is also proposed to increase the prediction accuracy. Case studies are performed to validate the effectiveness of the proposed models. An average of 4.25% error is observed for the AM case, which is significantly smaller than the prediction error of the existing models in the literature. Furthermore, in the hybrid case, an average of 91.83% accuracy is obtained.