Search

A major and stable QTL cQSGR.sau.3D, which can explain 33.25% of the phenotypic variation in SGR, was mapped and validated, and cQSGR.sau.3D was found to be independent of GI. In this study, a recombinant inbred line (RIL) population containing 304 lines derived from the cross of Chuan-nong17 (CN17) and Chuan-nong11 (CN11) was genotyped using the Wheat55K single-nucleotide polymorphism array. A high-density genetic map consisting of 8329 markers spanning 4131.54 cM and distributed across 21 wheat chromosomes was constructed. QTLs for whole spike germination rate (SGR) were identified in multiple years. Six and fourteen QTLs were identified using the Inclusive Composite Interval Mapping-Biparental Populations and Multi-Environment Trial methods, respectively. A total of 106 digenic epistatic QTLs were also detected in this study. One of the additive QTLs, cQSGR.sau.3D, which was mapped in the region from 3.5 to 4.5 cM from linkage group 3D-2 on chromosome 3D, can explain 33.25% of the phenotypic variation in SGR and be considered a major and stable QTL for SGR. This QTL was independent of the seeds' germination traits, such as germination index. One Kompetitive Allele-Specific PCR (KASP) marker, KASP-AX-110772653, which is tightly linked to cQSGR.sau.3D, was developed. The genetic effect of cQSGR.sau.3D on SGR in the RIL and natural populations was successfully confirmed. Furthermore, within the interval in which cQSGR.sau.3D is located in Chinese Spring reference genomes, thirty-seven genes were found. cQSGR.sau.3D may provide new resources for pre-harvest sprouting resistance breeding of wheat in the future.

Graph Neural Networks (GNN) has been used to exploit global features in text representation learning for natural language processing (NLP) tasks, including text classification, sequence tagging, neural machine translation and relational reasoning. However, GNN based models usually build a graph for the entire corpus, they have high memory consumption, ignoring the order of words and containing test documents in the training graph. Thus, these models are inherently transductive and have difficulties in inductive learning. In order to solve the above problems, we propose a Graph Transformer Networks based Text representation (GTNT) model. It first constructs a degree-centric text graph, which generates a text graph for each document in the corpus. Then it adopts a graph transformer network to model the graph to obtain node embeddings. When we apply our proposed GTNT model to citation recommendation and text classification tasks, the experimental results show that our model outperforms other state-of-the-art models.

This paper presents a new 3D heterogeneous integration scheme for inertial Micro-system, where a TSV interposer with a cavity is used to separate two aligned TSV interposers that are mounted with inertial MEMS (Micro Electro-Mechanical System) devices, ASIC (Application Specific Integrated Circuit) chips, and passive components receptively. A finite element simulation method that uses equivalent elements and sub-model methods is developed to do design for reliability in Thermal Cycling Test (TCT). Comparing with the normal simulation method, an agreement is obtained in the stress distribution in the MEMS chips, TSVs and micro bumps. In the meantime, the simplified simulation method provides a competitive accuracy, the difference is 0.88% in the maximum deformation of the whole model between the two methods, the difference is 0.41% in the maximum deformation of the TSV and 0.65% in the micro bump at the most dangerous position. Base on this simplified simulation method, a 3D heterogeneous inertial micro-system is analyzed and optimized. The design of ladder TSV with the usage of underfill and gold balls is selected. The simulation results also showed that RDL (Redistribution Layer) can make significant influence on the stress distribution in the systems. If considering RDL, the maximum stress inside the micro bump at the most dangerous location increases by 62.79% while the max stress in TSV decreases by 14.03%, which means that stress optimization should also be carried out after designing the RDL layout.

Drug repurposing refers to discovery of new medical instructions for existing chemical drugs, which has great pharmaceutical significance. Recently, large-scale biological datasets are increasingly available, and many graph neural network (GNN) based methods for drug repurposing have been developed. These methods often deem drug repurposing as a link prediction problem, which mines features of biological data to identify drug-disease associations (i.e., drug-disease links). Due to heterogeneity of data, we need to deeply explore heterogeneous information of biological network for drug repurposing. In this paper, we propose a Relation-aware Heterogeneous Graph Transformer (RHGT) model to capture heterogeneous information for drug repurposing. We first construct a drug-gene-disease interactive network-based on biological data, and then propose a three-level network embedding model, which learns network embeddings at fine-grained subtype-level, node-level and coarse-grained edge-level, respectively. The output of subtype-level is the input of node-level and edge-level, and the output of node-level is the input of edge level. We get edge embeddings at edge-level, which integrates edge type embeddings and node embeddings. We deem that in this way, characteristics of drug-gene-disease interactive network can be captured more comprehensively. Finally, we identify drug-disease associations (i.e., drug-disease links) based on the relationship between drug-gene edge embeddings and gene-disease edge embeddings. Experimental results show that our model performs better than other state-of-the-art graph neural network methods, which validates effectiveness of the proposed model.

Unmanned Aerial Vehicles (UAVs) have been widely employed in the Internet of Things (IoT) networks due to their high mobility and high probability of line-of-sight (LoS) propagation. Equipped with certain payloads, UAVs are able to gather data from sensors located in a particular area where no ground base station is available for transmitting data, such as oceans and mountains. However, for a time-sensitive network, the latency has to be minimized, especially in heterogeneous scenarios where each sensor has its own latency tolerance, which emphasizes the importance of trajectory design of UAVs. In this paper, we propose a priority-oriented trajectory planning problem for a UAV-aided time-sensitive heterogeneous IoT network, based on which we provide a solution for satisfying the latency tolerance of the network within a given period of time. Aiming at optimizing trajectories, we employ continuous Deep Q-Learning Network (DQN) which is proven to be capable of identifying a relatively optimal trajectory compared to the benchmarks through a large number of experiments. Simulation results are provided for demonstrating that the proposed DQN-based algorithm outperforms the benchmarks. More particularly, the proposed DQN-based algorithm is capable of achieving in excess of 49% and 10% improvements in system costs over the greedy algorithm and Q-Learning algorithm, respectively.

Heterogeneous information network (HIN), which contains various types of nodes and links, has been applied in recommender systems. Although HIN-based recommendation approaches perform better than the traditional recommendation approaches, they still have the following problems: for example, meta-paths are manually selected, not automatically; meta-path representations are rarely explicitly learned; and the global and local information of each node in HIN has not been simultaneously explored. To solve the above deficiencies, we propose a tri-attention neural network (TANN) model for recommendation task. The proposed TANN model applies the stud genetic algorithm to automatically select meta-paths at first. Then, it learns global and local representations of each node, as well as the representations of meta-paths existing in HIN. After that, a tri-attention mechanism is proposed to enhance the mutual influence among users, items, and their related meta-paths. Finally, the encoded interaction information among the user, the item, and their related meta-paths, which contain more semantic information can be used for recommendation task. Extensive experiments on the Douban Movie, MovieLens, and Yelp datasets have demonstrated the outstanding performance of the proposed approach.

Automatic modulation classification (AMC) technology, which utilizes to classify different kinds of signals using various expert features, typically the constellation map and the cyclic spectrum density graph, plays a significant role in spectrum monitoring and radio communication. However, due to their own disadvantages of constellation and cyclic spectrum density when recognizing some certain signals, it is necessary to combine them together in AMC. In this paper aiming at the commonly used features-based (FB) approach in practice, we propose a novel AMC model which jointly utilizes constellation map and cyclic spectrum density graph as the signal features. In order to provide a solution for the model, we propose a feature-based and supervised network in Single in Single out (SISO) system to identify seven kinds of signals, which is called Feature Fusion Convolutional Network (FFCN). By employing modified residual neural network (ResNet), the network can accomplish the prediction and classification according to the normalized and combined input from both two approaches. Simulation results are provided to show that FFCN can classify different types of signals with high probability as well as maintain the splendid efficiency compared to the benchmarks. Typically when signal to noise ratio (SNR) is low, the success rate of the proposed network is 2% higher at average, which proves that our network has better performance.

Gait, the walking pattern of individuals, is one of the most important biometrics modalities. Most of the existing gait recognition methods take silhouettes or articulated body models as the gait features. These methods suffer from degraded recognition performance when handling confounding variables, such as clothing, carrying and view angle. To remedy this issue, we propose a novel AutoEncoder framework to explicitly disentangle pose and appearance features from RGB imagery and the LSTM-based integration of pose features over time produces the gait feature. In addition, we collect a Frontal-View Gait (FVG) dataset to focus on gait recognition from frontal-view walking, which is a challenging problem since it contains minimal gait cues compared to other views. FVG also includes other important variations, e.g., walking speed, carrying, and clothing. With extensive experiments on CASIA-B, USF and FVG datasets, our method demonstrates superior performance to the state of the arts quantitatively, the ability of feature disentanglement qualitatively, and promising computational efficiency., To appear at CVPR 2019 as an oral presentation

Air-to-ground (A2G) communication is envisioned to support numerous applications in 5G wireless networks. In this paper, an active measurement campaign for unmanned aerial vehicle (UAV) channels is introduced. Simulation based on Graph Model (GM) is also conducted for this specific measurement. Both the measured and simulated channel impulse responses (CIRs) are extracted. Based on the CIRs, the multipath components (MPCs) are estimated by using a high-resolution algorithm derived according to the space-alternating generalized expectation-maximization (SAGE) principle. The GM simulation performance is assessed by comparing the concatenated power delay profiles (CPDPs) and MPCs with the measurement results. In addition, an object-labeled clustering method is proposed to further analyze the low altitude UAV channel contributed by individual objects/buildings.

Low altitude unmanned aerial vehicle (UAV)-aided applications are promising in the future generation communication systems. In this paper, a recently conducted measurement campaign for characterizing the low-altitude air-to-ground (A2G) channel in a typical Long Term Evolution (LTE) network is introduced. Five horizontal flights at the heights of 15, 30, 50, 75, and 100 m are applied, respectively. The realtime LTE downlink signal is recorded by using the Universal Software Radio Peripheral (USRP)-based channel sounder onboard the UAV. Channel impulse responses (CIRs) are extracted from the cell specific signals in the recorded downlink data. To shed lights on the physical propagation mechanisms, propagation graph simulation is exploited. Moreover, path loss at different heights are investigated and compared based on the empirical data. The simulated and empirical results provide valuable understanding of the low altitude A2G channels.

Assessing user experience on a new vehicle design objectively is critical for an automotive company to improve customer satisfactory of its products. The technology of motion capture and digital human modeling has been utilized in many recent studies to assist in the design process of a new product. Motions of human interacting with a new product are captured or simulated. The human body trajectory and swept volumes of the motion are overlaid with the product geometry in CAD, providing a guidance for the direction of new design changes. However, a CAD system generally requires some training to be efficiently used to review a design. A more intuitive and easier approach is preferred. In recent years, Virtual and Augmented Reality (VR/AR) are taking root in engineering area. The technology provide great advantages to the applications of design review, process simulation, maintenance, and training, etc., by expanding the physical world with virtual components or helpful information. This paper presents a new method of motion-based immersive design system for vehicle occupant package that integrates the technologies of VR/AR and digital human modeling. The method incorporates human motions captured or simulated along with the new geometry of a vehicle compartment inside a VR/AR environment. Through this, the designer is able to not only immersively experience the new design, but also observe the behavior of different users interacting with the design in the 3D environment. It provides a systematic and intuitive approach for a designer to quickly iterate through the test-review-revise design cycle and achieve a more accommodating and occupant friendly vehicle compartment design.

Most occupant accommodation assessments of a new vehicle design currently still utilize human appraisal. That is, human subjects experience the new design physically and provide feedback including a numerical rating or verbatim description. There are two drawbacks with this type of assessment: 1) the outcome is subjective. They are likely affected by other factors such as the vehicle’s appearance or brand and the individual’s own bias; 2) the outcome may not be able to reveal where the issues are nor how to resolve them. The digital manikin technology has been widely used in different areas: starting from movie and video gaming industries, and getting more and more involved in the product life circle of manufacture industries. Human motions are captured, and the digital manikin is utilized to present these motions virtually. This paper introduces a method that uses digital manikins to assist the process of vehicle design. Subjects’ motions interacting with a vehicle, which are related to a new design change, are captured. These motions are used to drive digital manikins that represent their respective subjects in size and body shape. A software system that animates the digital manikin according to the motions and creates swept volumes of selected body segments was created. The collection of the swept volumes of all subjects represents the space that is occupied by the human body during the motion. This space can be used to assess the design changes by indicating the minimum clearance between the swept volume and vehicle components or the interference between the human body with the components. In addition, the space described by the swept volumes provides a guideline or space limit for any future design changes. This method is objective. It not only pin-points the locations that cause discomfort or inconvenience by the new design, but also provides quantitative suggestions on how much improvement is needed for a better design.

One of the common usages of a captured human body motion in automotive application is creating swept volumes of the body surfaces based on the trajectories of its motion. Recent development of depth sensors enables fast and natural motion capture without attaching markers on subjects’ bodies. Microsoft Kinect is one of widely used depth sensors. It can track a whole body motion and output a skeleton model. A new method is developed to create the swept volumes from the motion captured by Kinect using an open source graphic system. The skeleton motion is recorded in a file format that is flexible to retain the skeleton’s structure and acceptable to various graphic systems. The motion is then bound with a surface manikin model in the graphic system, where the swept volumes are generated. This method is more flexible and portable than utilizing a commercial digital manikin, and potentially provides more accurate result.

In today's competitive market, it is essential for companies to provide products which not only achieve high performance, but also appeal to the tastes of consumers. Therefore, a key element of des...

The vehicle body-in-prime static stiffness can significantly affect vehicle NVH, ride and handling, and durability performance. Stiffness target setting is a crucial step in a vehicle design and development process. Traditionally, benchmarking data and CAE models are used to help identify and set the target, yet they have proven to be time consuming and very subjective. On the other hand, statistical models obtained through data mining or meta-modeling have been tested and applied to various problems, including design optimization. In the present paper, we attempt to utilize this technology for providing more objective vehicle static stiffness targets. Stiffness data characteristics of various vehicle classes are benchmarked. Different statistical modeling techniques are examined using the data to set stiffness targets. The validity and limitations of these techniques are also discussed.

The ease of entering a car is one of the important ergonomic factors that car manufacturers consider during the process of car design. This has motivated many researchers to investigate factors that affect discomfort during ingress. The patterns of motion during ingress may be related to discomfort, but the analysis of motion is challenging. In this paper, a modeling framework is proposed to use the motions of body landmarks to predict subjectively reported discomfort during ingress. Foot trajectories are used to identify a set of trials with a consistent right-leg-first strategy. The trajectories from 20 landmarks on the limbs and torso are parameterized using B-spline basis functions. Two group selection methods, group nonnegative garrote (GNNG) and stepwise group selection (SGS), are used to filter and identify the trajectories that are important for prediction. Finally, a classification and prediction model is built using support vector machine (SVM). The performance of the proposed framework is then evaluated against simpler, more common prediction models.Copyright © 2014 by ASME

The ease of getting in and out of a vehicle (or ingress/egress) is one of the most important ergonomic issues for automotive manufacturers. It represents the first physical contact of a customer with a vehicle. A pleasant sensation while interfacing with the vehicle plays a vital role in vehicle purchasing decisions. Understanding and being able to assess vehicle ingress/egress performance early in a design process is therefore critical to a successful vehicle design. Conventional method relies on clinic research with physical bucks, which is a time consuming and pure subjective process. A new hardware-in-the-loop motion analysis system has been developed, which uses the latest motion capture, biomechanical and digital human modelling technologies to capture and analyse human motions as the driver or passenger interacts with a vehicle. The design assessment is provided in both subjective ratings and, for the first time, the physical data (e.g., swept volumes in CAD). The use of the system avoids costly seating buck builds and reduces engineering time and cost associated with both buck build and conducting the tests. Most importantly, it enables engineers to assess a vehicle’s ingress/egress performances early in its design process, which will lead to better vehicle packages and better customer satisfaction.

Choice models play a critical role in enterprise-driven design by providing a link between engineering design attributes and customer preferences. However, existing approaches do not sufficiently capture heterogeneous consumer preferences nor address the needs of complex design artifacts, which typically consist of many subsystems and components. An integrated Bayesian hierarchical choice modeling (IBHCM) approach is developed in this work, which provides an integrated solution procedure and a highly flexible choice modeling approach for complex system design. The hierarchical choice modeling framework utilizes multiple model levels corresponding to the complex system hierarchy to create a link between qualitative attributes considered by consumers when selecting a product and quantitative attributes used for engineering design. To capture heterogeneous and stochastic consumer preferences, the mixed logit choice model is used to predict consumer system-level choices, and the random-effects ordered logit model is used to model consumer evaluations of system and subsystem level design features. In the proposed approach, both systematic and random consumer heterogeneity are explicitly considered, the ability to combine multiple sources of data for model estimation and updating is provided using the Bayesian estimation methodology, and an integrated estimation procedure is introduced to mitigate error propagated throughout the model hierarchy. The new modeling approach is validated using several metrics and validation techniques for behavior models. The benefits of the IBHCM method are demonstrated in the design of an automobile occupant package.

Human appraisals are becoming increasingly important in the design of engineering systems to link engineering design attributes to customer preferences. Human appraisals are used to assess consumers’ opinions of a given product design, and are unique in that the experiment response is a function of both the product attributes and the respondents’ demographic attributes. The design of a human appraisal is characterized as a split-plot design, in which the respondent demographic attributes form the whole-plot factors while the product attributes form the split-plot factors. The experiments are also characterized by random block effects, in which the design configurations evaluated by a single respondent form a block. An experimental design algorithm is needed for human appraisal experiments because standard experimental designs often do not meet the needs of these experiments. In this work, an algorithmic approach to identify the optimal design for a human appraisal experiment is developed, which considers the effects of respondent fatigue and the block and split-plot structure of such a design. The developed algorithm seeks to identify the experimental design which maximizes the determinant of the Fisher Information Matrix, labeled as the D-criterion of a given design. The algorithm is derived assuming an ordered logit model will be used to model the rating responses. The advantages of this approach over competing approaches for minimizing the number of appraisal experiments and model-building efficiency are demonstrated using an automotive interior package human appraisal as an example.Copyright © 2008 by ASME

Choice models play a critical role in enterprise-driven design by providing a link between engineering design attributes and customer preferences. In our previous work, we introduced the hierarchical choice modeling approach to address the special needs in complex engineering system design. The hierarchical choice modeling approach utilizes multiple model levels to create a link between qualitative attributes considered by consumers when selecting a product and quantitative attributes used for engineering design. In this work, the approach is expanded to the Bayesian Hierarchical Choice Modeling (BHCM) framework, estimated using an All-at-Once (AAO) solution procedure. This new framework addresses the shortcomings of the previous method while providing a highly flexible modeling framework to address the needs of complex system design. In this framework, both systematic and random consumer heterogeneity is explicitly considered, the ability to combine multiple sources of data for model estimation and updating is significantly expanded, and a method to mitigate error propagated throughout the model hierarchy is developed. In addition to developing the new choice model approach, the importance of including a complete representation of consumer heterogeneity in the model framework is provided. The new modeling framework is validated using several metrics and techniques. The benefits of the BHCM method are demonstrated in the design of an automobile occupant package.

The automation of tolerance analysis of assemblies is based on the development of the data structure for their representation in geometric modeling, and the computerization of procedures developed by Bjorke. The data structure contains assembly-level and component-level relationships to automate the generation of the tolerance chain for a user-selected sum dimension. A statistical analysis is performed on a sum dimension, the characteristics of which can be easily generated by the hierarchical relationships and the mating conditions between components contained within the assemblies' geometric data structure. The method uses a truncated unit beta distribution for the sum dimension and the contributions to the sum dimension by the statistical properties of 15 link types. The user supplies the tolerance range of the sum dimension and its location within the distribution. The program provides the user with a graphical display of the sum-dimension distribution, the specified range and location of the tolerance, and the results of the numerical integration of the distribution that generates the confidence limits for a given application.

This paper presents a scheme for representing features, dimensions, and tolerances in geometric models. The scheme combines a modified CSG tree with a face-edge based B-rep data structure to represent high-level form features and low-level primitive features. Dimensions and tolerances are treated as constraints. The dimensional relations within a machine part are described by constraining operators from which the position and orientation of form features are derived automatically. The modification of geometry can be achieved by directly changing dimensions. The representation promises a new engineering tool for design and manufacturing engineers and provides an important foundation for supporting engineering applications such as tolerance analysis and dimensioning checking. The scheme has been implemented in a prototype system.

A scheme of representing assemblies and an algorithm for the tolerance chain generation are developed so that tolerance chains of assemblies can be generated automatically to accommodate tolerance analysis at the assembly level. In the hierarchical data structure representing an assembly the connectivity information is carried by the instances of components and subassemblies, and the mating relations between each pair of mating entities are described by mating links, mating paths, mating conditions, and mating features. Mating graphs are derived from the mating links before they are searched for the generation of tolerance chains. The scheme has been implemented in a prototype interactive package that allows the user to model assemblies directly without detailed object modeling. Several examples of various complexity have been tested with success.

Demand models play a critical role in enterprise-driven design by expressing revenues and costs as functions of product attributes. However, existing demand modeling approaches in the design literature do not sufficiently address the unique issues that arise when complex systems are being considered. Current approaches typically consider customer preferences for only quantitative product characteristics and do not offer a methodology to incorporate customer preference-data from multiple component/subsystem-specific surveys to make product-level design trade-offs. In this paper, we propose a hierarchical choice modeling approach that addresses the special needs of complex engineering systems. The approach incorporates the use of qualitative attributes and provides a framework for pooling data from multiple sources. Heterogeneity in the market and in customer-preferences is explicitly considered in the choice model to accurately reflect choice behavior. Ordered logistic regression is introduced to model survey-ratings and is shown to be free of the deficiencies associated with competing techniques, and a Nested Logit-based approach is proposed to estimate a system-level demand model by pooling data from multiple component/subsystem-specific surveys. The design of the automotive vehicle occupant package is used to demonstrate the proposed approach and the impact of both packaging design decisions and customer demographics upon vehicle choice are investigated. The focus of this paper is on demonstrating the demand (choice) modeling aspects of the approach rather than on the vehicle package design.Copyright © 2007 by ASME

Appraising vehicle package design concepts using seating bucks — physical prototypes representing vehicle package, is an integral part of the vehicle package design process. Building such bucks is costly and may impose substantial burden on the vehicle design cycle time. Further, static seating bucks lack the flexibility to accommodate design iterations during the gradual progression of a vehicle program. A “Computer controlled seating buck”, as described in this paper, is a quick and inexpensive alternative to the traditional seating bucks with the desired degree of fidelity. It is particularly useful to perform package and ergonomic studies in the early stages of a vehicle program, long before the data is available to build a traditional seating buck. Such a seating buck has been developed to accommodate Ford vehicle package design needs. This paper presents the functional requirements, the high level conceptual design of how these requirements are realized, and the methods to verify, improve and sustain the dimensional accuracy and capability of the new computer controlled seating buck.Copyright © 2006 by ASME

Early conceptual design is one of the most important stages in vehicle product development. At this stage, vehicle design information is limited. In many cases, historical or benchmark vehicle data are used as surrogates in decision making. Some of the legacy components may be reused in the new vehicle design in order to reduce development and manufacturing costs. Nowadays, parametric modeling methods have been developed and employed for conceptual vehicle design. However, how to quickly reuse the previous design and legacy vehicle data in parametric design is still a challenge. A hybrid method for parametric conceptual vehicle design is presented in this paper, which uses both legacy components and parametric surfaces. This method allows easy reuse of historical data together with new parametric surfaces for early vehicle design. It enables mix and match of legacy components with newly designed parametric surfaces in representing a new design. This method provides a systematic way to enforce the commonality and reusability in a vehicle design.Copyright © 2006 by ASME