Search

This paper presents a low-cost, accurate indoor positioning system that integrates image acquisition and processing and data-driven modeling algorithms for robotics research and education. Multiple overhead cameras are used to obtain normalized image coordinates of ArUco markers, and a new procedure is developed to convert them to the camera coordinate frame. Various data-driven models are proposed to establish a mapping relationship between the camera and the world coordinates. One hundred fifty data pairs in the camera and world coordinates are generated by measuring the ArUco marker at different locations and then used to train and test the data-driven models. With the model, the world coordinate values of the ArUco marker and its robot carrier can be determined in real time. Through comparison, it is found that a straightforward polynomial regression outperforms the other methods and achieves a positioning accuracy of about 1.5 cm. Experiments are also carried out to evaluate its feasibility for use in robot control. The developed system (both hardware and algorithms) is shared as an open source and is anticipated to contribute to robotic studies and education in resource-limited environments and underdeveloped regions.

Textured microstructure of B- and P- doped Ni-rich lithium layered oxide is inherited from the hydroxide precursor morphology and the mechanism is related to the formation of a lithium-containing amorphous layer on the surface of primary particles.

The insulator-to-metal phase transition from VO2(M1) to VO2(R) results in extensive improvements in electrocatalytic activity for both the OER and the HER.

The substitutional solid solution Mn1-xVxP compounds are proposed as a high performance anode for lithium ion batteries (LIBs) through a novel alloying/insertion hybrid electrochemical reaction concept by combining alloying reaction-type MnP and insertion reaction-type VP. The solid solution series of Mn1-xVxP are successfully synthesized via a facile high energy mechanical milling. Their electrochemical properties as an anode for LIBs are systematically studied and compared with those of MnP/VP mixture, particularly focusing on the verification of simultaneous alloying/insertion hybrid electrochemical reaction in the solid solution compounds. The Mn0.75V0.25P solid solution electrode shows the excellent high rate cyclability delivering the reversible capacity of 321 mAh g−1 after 5000 cycles at a high current density of 1.0 A g−1, resulting from the synergistic effects of two reaction mechanisms. The homogeneously substituted vanadium ions enable the alloying/insertion hybrid electrochemical reaction in a Mn1-xVxP single phase, which can effectively reduce the rate of volume change, hinder the pulverization and agglomeration of alloying reaction-type Li-Mn-P crystallite during cycling, and ensure the fast electron and ion transport. This simple, yet innovative, solid solution design with the consideration of structural relationships and electrochemical properties inspires the development of advanced ternary or multi-component compound electrode materials for LIBs or other energy storage devices.

Powder metallurgy (P/M) has been widely used in automobile, home appliances, and electronic devices, but its uses are very limited due to the low relative density of 85 to 95 pct. The Fe–Ni–B–C alloy system can mitigate the aforementioned issues by the liquid phase sintering, which results in a nearly full densification. However, the boron-containing alloys produced the brittle eutectic phases [Fe3(C, B) and Fe2B] along the grain boundaries, which are detrimental to the mechanical properties. The main objective of this study is to improve the ductility of boron-containing alloys through the microstructure modification. For this, the volume fraction of solidified phase was optimized by controlling the composition, and the coarsening of solidified α-Fe particles into the pearlite matrix was induced by a post annealing, which reduces the continuous network of eutectic phases and increased the grain continuity. In addition, the effect of microstructure modification on the mechanical properties of Fe–B–C and Fe–Ni–B–C alloys was comparatively investigated. As a result of microstructure modification, the post-annealed Fe–1Ni–0.4B–0.8C alloy exhibited a high elongation to failure of 5.2 pct.

This study reports a facile method to synthesize the calcium silicate-based cement with an enhanced radiopacifying performance. Sr-substituted Ca3SiO5 (Ca3−xSrxSiO5, x = 0.0 ~ 0.2) was prepared by solid-state reaction, and its phase, hydration behavior, and radiopacity were investigated and compared with those of 10 wt% Bi2O3-added Ca3SiO5. A nearly single-phase Ca3SiO5 was synthesized at 1400 °C and the solubility limit of Sr into Ca3SiO5 was approximately 3.3 mol % (x = 0.1). The Sr substitution did not affect the size and morphology of Ca3SiO5 powder. The extent of hydration was slightly delayed in Sr-substituted Ca3SiO5, but the hydration products were the same composing of fiber-like calcium silicate hydrate (C–S–H) and plate-like calcium hydroxide (Ca(OH)2). The radiopacity of Ca3SiO5 was significantly enhanced with Sr substitution and it was slightly higher than that of Bi2O3-added Ca3SiO5 at a similar concentration. Consequently, Sr-substituted Ca3SiO5 can be a promising dental root canal filling material with an improved radiopacifying property.

2-Dimensional semiconductor SnS2 has emerged as one of the most prospective candidates for chemoresistive gas sensor applications due to its outstanding gas sensing performance. Herein, we propose the room temperature nitrogen dioxide (NO2) sensing of SnS2 nanoflowers (NFs) enabled by visible light activation. SnS2 NFs were successfully prepared by solvothermal synthesis with abundant edge sites. The high absorbance in the visible light region triggered the generation of charge carriers resulting in decreased resistance and enhanced gas sensing characteristics. Even under red light and green light with low photon energy, the room temperature NO2 sensing performance was improved. The highest NO2 sensing performance was accomplished under blue light, including the highest response, excellent selectivity towards NO2 and an extremely low detection limit. Moreover, the sensor exhibited reliable gas sensing performance in humid conditions and maintained its properties after long-term relaxation. Taking advantage of surface properties, optical properties and gas sensing properties, a light-activated SnS2 NF based gas sensor is anticipated to further develop SnS2 nanostructures for use in an electronic nose.

The realization of high performance Ni-rich layered cathodes remains a challenge because of the multiple degradation factors that concurrently operate during battery cycling. In particular, depletion of oxygen charge and consequent lattice-oxygen instability at deep charge state accelerate the subsequent chemomechanical degradation mechanisms. Among the proposed methodologies, doping has proven to be effective in enhancing the cathode cycle life by stabilizing the layered structure. Herein, we achieved the electrochemically stabilized Ni-rich LiNi0.92Co0.04Mn0.04O2 through Zr doping, resulting in a 15% increase of the capacity retention after 100 cycles. In-depth investigations are conducted to unveil the effects of Zr doping on the layered cathode, and in particular, the critical role of Zr doping on the lattice oxygen stability is systematically studied. By combining state-of-the-art magnetometer characterization, X-ray analysis, and first-principles calculation, we reveal that Zr doping positively contributes the to lattice oxygen stability by alleviating the oxygen charge loss at deep charge, thereby improving the cathode electrochemical reversibility. Our findings provide an insight into the Zr doping mechanism and help to design Ni-rich layered oxides for future applications.

High energy density lithium ion batteries (LIBs) are in urgent demand for portable electronic devices and electrical vehicles. As the energy density heavily relies on the cathode materials, extensive researches have been undertaken to develop the cathode materials with a high degree of lithium utilization. In this respect, the layered lithium transition metal oxides with the composition of LiNixCoyMn1−x−yO2 (NCM) have received much attention due to their low cost and high practical capacity. Unfortunately, the viability of NCM for the market is hindered by the challenges resulting from the deteriorations occurring on the length scale from nanometer to bulk. These degradation mechanisms interact each other during battery cycling and simultaneously deteriorate the NCM lifespan. Therefore, this review article focuses on the insightful understanding of the capacity/voltage fading mechanisms of NCM with an aim to provide important clues for the future design of layered lithium transition metal oxides. First, we discuss the crystal and electronic structure of the layered oxide as they are related to the NCM mechanochemical breakdown. Then, on overviewing each degradation mechanism, we provide an insightful opinion with regard to finding the correlation between the mechanisms. Finally, a variety of novel approaches that have been proposed with an attempt to prolong the cycle life of NCM cathodes are presented.

The a disintegrin and metalloprotease (ADAM) family proteins comprise a group of membrane-anchored proteins. ADAM32 is expressed specifically in testis and is closely related phylogenetically to ADAM2 and ADAM3, which are known to be critical for fertilization in mice. To assess the biological role of ADAM32, we analyzed Adam32-mutant mice. We found that male mice lacking ADAM32 have normal fertility, testicular integrity, and sperm characteristics. ADAM32 was found to exist at lower levels than ADAM2 and ADAM3 in wild-type testis and sperm, respectively. The present study demonstrates that ADAM32 is dispensable for fertility and appears to be functionally unrelated to ADAM2 and ADAM3 in mice.

This paper presents a surrogate- and possibility-based design optimization (SPBDO) methodology for robust design of convective PCR. Parametric computational fluid dynamics (CFD) models are built and simulated at sampled locations within the design space to capture the effect of design configurations on thermofluidic transport and convection–diffusion-reaction in the convective PCR. A support vector machine-based classifier model is trained to retain only practically relevant data for enhanced surrogate modeling accuracy. Surrogate models are constructed by Kriging interpolation and multivariate polynomial regression methods to establish the mapping between design configurations and DNA doubling time (indicative of reactor performance). Then a process to combine the sequential method of PBDO and the surrogate model is developed, and a trade study is carried out to evaluate the impact of possibility of failure ( $$\alpha$$ -value) and the balance between performance and design reliability. Our study demonstrates that the proposed SPBDO represents an effective method to consider robustness in PCR design for POC applications, especially when the uncertainty information or possibilistic characteristics of design variables is limited.

A MnV2O6/graphene nanocomposite was fabricated through hydrothermal synthesis and high energy milling to introduce it as an efficient OER electrocatalyst. The MnV2O6/graphene nanocomposite with 20 wt% graphene exhibited superior electrocatalytic OER performance with a low overpotential and high stability and durability in 1 M KOH aqueous solution, exhibiting even after 1000 CV cycles.

In the pandemic era, the development of high-performance indoor air quality monitoring sensors has become more critical than ever. NO

International audience; Heat shock factor 2 (HSF2) regulates the transcription of the male-specific region of the mouse Y chromosome long arm (MSYq) multicopy genes only in testes, but the molecular mechanism underlying this tissue specificity remains largely unknown. Here, we report that the testicular germ cell-specific long noncoding RNA (lncRNA), NR_038002, displays a characteristic spatiotemporal expression pattern in the nuclei of round and elongating spermatids. NR_038002-knockout male mice produced sperm with abnormal head morphology and exhibited reduced fertility accompanied by a female-biased sex ratio in offspring. Molecular analyses revealed that NR_038002 interacts with HSF2 and thereby activates expression of the MSYq genes. We designate NR_038002 as testicular germ cell-specific HSF2-interacting lncRNA (Teshl). Together, our study is the first to demonstrate that the testis specificity of HSF2 activity is regulated by the lncRNA Teshl and establishes a Teshl-HSF2-MSYq molecular axis for normal Y-bearing sperm qualities and consequent balanced offspring sex ratio.

In mammals, the early embryo travels down the oviduct to the uterus and prepares for implantation. The unique features of preimplantation development include compaction followed by blastocyst formation. This first cell lineage specification involves various proteins including cell polarity regulators, kinases, and transcription factors. In this study, a novel gene named predicted gene 11545 (Gm11545) expressed predominantly in mouse early embryos was identified and characterized at the transcript, protein, cellular, and functional levels. The Gm11545 protein localized to both cytoplasmic and membrane regions of preimplantation embryos. Remarkably, knockdown of Gm11545 led to arrest of mouse embryos at the morula stage and consequent impairment of blastocyst formation. Expression patterns of the key transcription factors critical for early lineage specification, octamer-binding transcription factor 4 and caudal type homeobox 2, were affected by Gm11545 depletion. Based on the collective findings, we propose that the novel protein identified in this study, Gm11545, is implicated in cell proliferation and cell lineage specification critical for blastocyst formation.-Kim, J., Kim, J., Jeong, J., Hong, S. H., Kim, D., Choi, S., Choi, I., Oh, J. S., Cho, C. Identification of a novel embryo-prevalent gene, Gm11545, involved in preimplantation embryogenesis in mice.

Silicon (Si) is considered to be one of the most promising anode candidates for next-generation lithium-ion batteries because of its high theoretical specific capacity and low discharge potential. However, its poor cyclability, caused by tremendous volume change during cycling, prevents commercial use of the Si anode. Herein, we demonstrate a high-performance Si anode produced via covalent bond formation between a commercially available Si nanopowder and a linear polymeric binder through an esterification reaction. For efficient ester bonding, polyacrylic acid, composed of -COOH groups, is selected as the binder, Si is treated with piranha solution to produce abundant -OH groups on its surface, and sodium hypophosphite is employed as a catalyst. The as-fabricated electrode exhibits excellent high rate capability and long cycle stability, delivering a high capacity of 1500 mA h g

Epididymal maturation is critical for acquisition of motility and fertilizing capacity by sperm. During epididymal transit, the surface of sperm undergoes prominent sequential changes through interactions with secreted proteins, including protease inhibitors. In the present study, we characterized three epididymis-specific SPINKs (serine protease inhibitors, Kazal-type): SPINK8, SPINK11, and SPINK12. We found that these epididymal SPINKs are expressed in an epididymal region-specific manner and their expression is developmentally regulated. Remarkably, cellular analyses revealed that SPINK8 and SPINK12 are transferred to the sperm. To investigate the in vivo properties of SPINK12, we analyzed knockout mice generated by CRISPR/Cas9-mediated genome editing. Loss of SPINK12 did not alter epididymal tubule structure or sperm phenotypes. Spink12 mutant mice exhibited normal fertility, suggesting that SPINK12 is functionally redundant in the epididymis.

To meet the ever growing demand for high energy density and high power density lithium ion batteries, silicon (Si) has been spotlighted as a promising anode material due to its high theoretical specific capacity. However, the practical use of Si anodes is hindered by the large volume change during the lithiation/delithiation process and resultant electrode pulverization. Significant progress has been made using nanostructured Si-based materials, but their low initial coulombic efficiency and low volumetric capacity are undesirable for practical applications. Si microparticles (SiMPs) are considered as an alternative because of low cost, commercial availability, and high tap density. Herein, we introduce a novel concept of in situ formation of a graphene oxide (GO)–polyacrylic acid (PAA) composite cage encapsulating silicon microparticles (SiMPs) as an anode for LIBs. The mechanically robust covalent bond was formed between the carboxylic acid group of the PAA binder and the hydroxyl group of GO through an esterification reaction. The ester bonded GO–PAA cage confined the pulverized SiMPs within and maintained the structural integrity resulting in high coulombic efficiency, excellent capacity retention, and high rate capability. The as-fabricated electrode delivered a high capacity of 2300 mA h g−1 after 200 cycles at a high current density of 1000 mA g−1. Furthermore, stable cycling was achieved with a high Si mass loading (1.98 mA h cm−2 after 500 cycles) and in a full cell (1.45 mA h cm−2 after 100 cycles).

This paper evaluates the impact of residual acceleration noise on the estimation of the Earth’s time-varying gravity field for future low-low satellite-to-satellite tracking missions. The goal is to determine the maximum level of residual acceleration noise that does not adversely affect the estimation error. The Gravity Recovery And Climate Experiment (GRACE) has provided monthly average gravity field solutions in spherical harmonic coefficients for more than a decade. It provides information about land and ocean mass variations with a spatial resolution of ∼350 km and with an accuracy within 2 cm throughout the entire Earth. GRACE Follow-on was launched in May 2018 to advance the work of GRACE and to test a new laser ranging interferometer, which measures the range between the two satellites with higher precision than the K-Band ranging system used in GRACE. Moreover, there have been simulation studies that show, an additional pair of satellites in an inclined orbit increases the sampling frequency and reduces temporal aliasing errors. Given the fact that future missions will likely continue to use the low-low satellite-to-satellite tracking formation with laser ranging interferometry, it is expected that the residual acceleration noise will become one of the largest error contributor for the time-variable gravity field solution. We evaluate three different levels of residual acceleration noise based on demonstrated drag-free systems to find a suitable drag-free performance target for upcoming geodesy missions. We analyze both a single collinear polar pair and the optimal double collinear pair of drag-free satellites and assume the use of a laser ranging interferometer. A partitioned best linear unbiased estimator that was developed, incorporating several novel features from the ground up is used to compute the solutions in terms of spherical harmonics. It was found that the suitable residual acceleration noise level is around 2 × 10−12 ms−2 Hz−1/2. Decreasing the acceleration noise below this level did not result in more accurate gravity field solutions for the chosen mission architecture.

A new P2-type Na0.7(Ni0.6Co0.2Mn0.2)O2 was prepared via co-precipitation and its electrochemical properties as a cathode for sodium ion batteries were compared with those of O3-type Na(Ni0.6Co0.2Mn0.2)O2, focusing on phase stability and cycling performance. The P2-type delivered a high capacity of 108 mA h g-1 after 300 cycles at 2C.

The substitutional solid solution Mn1−xFexP compounds between alloying reaction-type MnP and conversion reaction-type FeP are successfully synthesized via facile high energy mechanical milling and their electrochemical properties as an anode for lithium ion batteries (LIBs) are investigated. A complete solid solution is formed between two end members and the Mn1−xFexP solid solution phosphide electrodes show an enhanced electrochemical performance, delivering a capacity of 360 mA h g−1 after 100 cycles at a high current density of 2 A g−1 when the advantages of the two reaction mechanisms are beneficially combined. These synergistic effects resulted from the in situ generated nanocomposite of the Li–Mn–P alloying element and the Fe nano-network in combination with the surrounding amorphous lithium phosphide, which effectively buffers the accompanying volume variation, hinders the aggregation of the alloying element, and ensures the electron and ion transport.

V4P7 nanoparticles were synthesized via high-energy mechanical milling and their electrochemical properties as an anode for sodium-ion batteries were studied and compared with those of VO2(B)/Na and V4P7/Li cells, focusing on the electrochemical reaction mechanism and cycle performance. The V4P7 showed excellent cycling behavior even without any conductive material.

This paper presents a framework for automated optimization of double-heater convective PCR (DH-cPCR) devices by developing a computational fluid dynamics (CFD) simulation database and artificial neural network (ANN) model. The optimization parameter space that includes the capillary tube geometries and the heater sizes of DH-cPCR is established, and a database consisting of nearly 10,000 CFD simulations is constructed. The database is then used to train a two-stage ANN models that select practically relevant data for modeling and predict PCR device performance. The trained ANN model is then combined with the gradient-based and the heuristics optimization approaches to search for optimal device configuration that possesses the shortest DNA doubling time. The entire design process including model meshing and configuration, parallel CFD computation, database organization, and ANN training and utilization is fully automated. Case studies confirm that the proposed framework can successfully find the optimal device configuration with an error of less than 0.3 s, and hence, representing a cost-effective and rapid solution of DH-cPCR device design.

Testis-specific lncRNA regulates expression of mouse Y chromosome genes and plays a critical role in offspring sex ratio., Heat shock factor 2 (HSF2) regulates the transcription of the male-specific region of the mouse Y chromosome long arm (MSYq) multicopy genes only in testes, but the molecular mechanism underlying this tissue specificity remains largely unknown. Here, we report that the testicular germ cell–specific long noncoding RNA (lncRNA), NR_038002, displays a characteristic spatiotemporal expression pattern in the nuclei of round and elongating spermatids. NR_038002-knockout male mice produced sperm with abnormal head morphology and exhibited reduced fertility accompanied by a female-biased sex ratio in offspring. Molecular analyses revealed that NR_038002 interacts with HSF2 and thereby activates expression of the MSYq genes. We designate NR_038002 as testicular germ cell–specific HSF2-interacting lncRNA (Teshl). Together, our study is the first to demonstrate that the testis specificity of HSF2 activity is regulated by the lncRNA Teshl and establishes a Teshl-HSF2-MSYq molecular axis for normal Y-bearing sperm qualities and consequent balanced offspring sex ratio.

This paper presents a new artificial neural network (ANN)-based system model that concatenates an optimized artificial neural network (OANN) and a neural network compensator (NNC) in series to capture temporally varying system dynamics caused by slow-paced degradation/anomaly. The OANN comprises a complex, fully connected multilayer perceptron, trained offline using nominal, anomaly free data, and remains unchanged during online operation. Its hyperparameters are selected using genetic algorithm-based meta-optimization. The compact NNC is updated continuously online using collected sensor data to capture the variations in system dynamics, rectify the OANN prediction, and eventually minimize the discrepancy between the OANN-predicted and actual response. The combined OANN–NNC model then reconfigures the model predictive control (MPC) online to alleviate disturbances. Through numerical simulation using an unmanned quadrotor as an example, the proposed model demonstrates salient capabilities to mitigate anomalies introduced to the system while maintaining control performance. We compare the OANN–NNC with other online modeling techniques (adaptive ANN and multinetwork model), showing it outperforms them in reference tracking of altitude control by at least 0.5 m and yaw control by 1 deg. Moreover, its robustness is confirmed by the MPC consistency regardless of anomaly presence, eliminating the need for additional model management during online operation.

Mammalian spermatogenesis is a highly organized process with successive mitotic, meiotic, and postmeiotic phases. This unique developmental process is characterized by the involvement of spermatogenic cell-specific genes. In this study, we identified and investigated testis expressed gene 13 (Tex13) family genes, consisting of Tex13a, Tex13b, Tex13c1, and Tex13d, in mice. All of these genes were transcribed specifically or predominantly in male germ cells, and their transcription was developmentally regulated. Proteins encoded by the Tex13 genes were predicted to have a conserved domain of ~ 145 amino acids. Tex13a, Tex13c1, and Tex13d encode additional C-terminal regions containing a short conserved sequence termed a zinc finger-RAN binding protein 2 (zf-RanBP2) or zf-RanBP2-like domain. As TEX13B reportedly has transcriptional repressor activity, we examined the effect of the TEX13 proteins on transcriptional regulation using a reporter assay. All of the TEX13 proteins exhibited transcriptional repressor activity. This activity was revealed to reside in the TEX13B-corresponding regions of TEX13A, TEX13C1, and TEX13D. Further, we found that the C-terminal regions of TEX13A, TEX13C1, and TEX13D also have inhibitory activities. These results suggest that male germ cell-specific or -predominant TEX13 proteins commonly function in transcriptional repression as transcription cofactors and/or RNA binding proteins.

Real-time model updating of active structures subject to unmodeled high-rate dynamic events require structural model updates on the timescale of 2 ms or less. Examples of active structures subjected to unmodeled high-rate dynamic events include hypersonic vehicles, active blast mitigation, and orbital infrastructure. Due to the unmodeled nature of the events of interest, the real-time model updating algorithm should circumvent any model pre-calculations. In this work, we present a methodology that updates the finite element analysis (FEA) model of a structure experiencing varying dynamics through online measurements. The algorithm is demonstrated for a testbed, comprised of a cantilever beam and a roller that serves as movable support. The structure’s state is updated (i.e. the position of the moving roller) by continuously updating the associated FEA model through an online adaptive meshing and search algorithm. The structure’s state is continuously estimated by comparing the measured signals with FEA models. New FEA models are built based on the enhanced estimate of the structure’s state through adaptive meshing for modal analysis and adaptive search space for the FEA model selection. The proposed methodology is verified experimentally in real-time using the testbed. It is demonstrated that the adaptive features can achieve accurate state estimations within the required 2 ms timescale.

Many researchers in the prognostics and health management community have begun exploring the use of deep neural networks for predicting remaining useful life (RUL) of mechanical systems. These models have consistently reestablished the state-of-the-art in RUL prediction performance on common benchmarks, such as the NASA C-MAPSS Aircraft Engine dataset. However, they do not attempt to capture the multiple sources of uncertainty that are inherent in their predictions. This paper presents an approach for estimating both epistemic and heteroscedastic aleatoric uncertainties that emerge in deep neural network models that are trained for RUL prediction and demonstrates that quantifying their overall impact on predictions can be extremely valuable in real-world systems, where decisions are sometimes made during uncertain operating conditions. First, a novel deep neural network architecture is proposed that demonstrates competitive performance on the NASA C-MAPSS FD001 and FD003 datasets. Then, this network is adapted to estimate epistemic and heteroscedastic aleatoric uncertainties in the RUL prediction problem. Finally, a study is carried out to observe the effects that augmenting the RUL truth data, i.e. utilizing piecewise linear truth curves in place of the actual truth data, have on the perceived uncertainties in the system. Case studies on the C-MAPSS FD001 dataset will show that utilizing the actual RUL truth data can yield more meaningful uncertainty estimates and more insight into the relationship between sensor data and an engine's time-to-failure.

This paper presents a new paradigm of deep neural network (DNN) for the inverse design of microfluidic concentration gradient generators (µCGGs) with complex network topology. In this method, a concentration gradient (CG) and design parameters yielding the CG are, respectively, used as inputs and outputs of DNN, and the relationship between them is mapped. Several new elements are also proposed, including utilization of fast-running physics-based component model in the closed form to generate a large amount of data for DNN learning which otherwise is not available through computationally demanding computational fluid dynamics (CFD) simulation; and a divide-and-conquer strategy and DNN formulation combining classification and regression to mitigate many-to-one design complications for enhanced accuracy. Several DNN structures are investigated and developed, including single fully connected neural network (FCNN), convolutional neural network, and a new cascade FCNN for a divide-and-conquer implementation. Case studies are performed on a triple-Y µCGG to evaluate design performance of the proposed method in a six-dimensional space that only includes sample concentrations at inlet reservoirs as design parameters, and in a nine-dimensional design space, to which inlet flow pressures are also added. It is verified in high-fidelity CFD simulation that widely used CGs can be produced using DNN-predicted design parameters accurately with average error