China
Africa
Explore chapters and articles related to this topic
Learning to Control Difficult Systems: Neural Nets
Published in Richard J. Jagacinski, John M. Flach, Control Theory for Humans, 2018
Richard J. Jagacinski, John M. Flach
An important aspect of the Jenkins and Yuhas approach to this control problem was to decompose the state space of variables that could potentially influence the tire angle. Namely, a possible description of the truck cab and trailer is in terms of the two-dimensional position of the back of the trailer, (xt, yt), the trailer heading angle, the cab heading angle, and the front tire angle. A control strategy can be represented as a surface in a five-dimensional space with each dimension corresponding to one of these variables. The surface specifies the appropriate front tire angle for each combination of the other four variables. The neural network implements a possible control surface, which is gradually adjusted by gradient descent to improve system performance. Higher dimensional surfaces are generally adjusted more slowly on the basis of performance feedback.
Choosing the Right People
Published in Nigel Seel, Business Strategies for the Next-Generation Network, 2006
Note that unlike Myers-Briggs and Keirsey, there are no types in the FFM. The five axes constitute a kind of five-dimensional personality space, and an individual taking an FFM personality assessment ends up with an aggregated score along each dimension, a personality vector in this space. In practice, each trait is further subdivided into six subtraits that are also chosen to minimize cross-correlation; the resulting raw personality vector has 30 components. The significance of this collection of numbers can be hard to understand—the typological analysis, once the letter codes are understood, seems easier to work with, which is probably why it is more popular in a corporate or clinical context. People who have used the FFM with clients have scored the assessment in terms of High, Median, or Low along each of the five main dimensions. This is a little more manageable, but you end up with 3^5 = 243 possible outcomes, rather than 16 types.
Multiaxial ratchetting for 2014-T6 aluminium alloy tube
Published in József Farkas, Károly Jármai, Tubular Structures VII, 1996
When a thin-walled tubular specimen is subjected to combined loading of axial force and torque, the corresponding strain and stress state may be conveniently expressed in terms of deviatonc vector planes of strain and stress tensors. The definition of the axial torsional subspace follows as a subspace of Ilyushin’s five-dimensional deviatoric vector space. The stress vector is defined as ()σ¯=σ1n¯1+σ3n¯3
A twin data and knowledge-driven intelligent process planning framework of aviation parts
Published in International Journal of Production Research, 2022
Jingjing Li, Guanghui Zhou, Chao Zhang
TDKIPP refers to a five-dimensional intelligent time-varying space (including physical space, virtual space, data space, knowledge space and decision-making space) process planning system that integrates technologies such as IoT, digital twin, data processing, knowledge service, process decision-making into manufacturing system from the perspective of data knowledge hybrid driving. It has the following characteristics: Digital twin model construction: TDKIPP studies the physical entity mechanism modelling based on the fusion of geometric quantities and multi-physical quantities, meanwhile, integrates processing data to construct a digital twin model, and covers the process decision-making, evaluation, feedback optimisation and other functions.Innovation methods of intelligent process planning based on digital twin: innovative methods are used for process planning, such as knowledge recommendation based on dynamic process knowledge base, independent intelligent process decision-making and evaluation, machining quality prediction and process feedback based on deep learning, migration learning, adaptive adjustment, which provide new ideas for future intelligent process planning.
Training a Neural-Network-Based Surrogate Model for Aerodynamic Optimisation Using a Gaussian Process
Published in International Journal of Computational Fluid Dynamics, 2022
Yousef Ghazi, Nahla Alhazmi, Radek Tezaur, Charbel Farhat
Figure 4 (top-right) visualises the points sampled in the five-dimensional parameter space. In this figure, the free-stream Mach number, free-stream angle of attack, and the camber are shown on the x, y, and z axes, respectively, the position of the maximum camber is represented by a varying colour, and the thickness by a varying size of the dots. A histogram of the distribution of each parameter over the sampled points is given in Figure 4 (bottom). The histograms reveal that the free-stream angle of attack and the camber have been sampled inside their intervals more finely than the other three parameters. The value of the true relative error (8) achieved after 88 parameter points have been adaptively sampled is 1.02%, which closely matches the value of the error indicator (7).
Digital twin in manufacturing: conceptual framework and case studies
Published in International Journal of Computer Integrated Manufacturing, 2022
Igiri Onaji, Divya Tiwari, Payam Soulatiantork, Boyang Song, Ashutosh Tiwari
Tao et al. (2018) presented a digital twin-driven product design (DTPD) framework. This serves as a guide on the creation of a product digital twin and the utilisation of its generated knowledge in the product design process. Zhang et al. (2019a) proposed a data and knowledge-driven digital twin framework for a manufacturing cell (DMTC). This supports an autonomous manufacturing cell using data for the perception of manufacturing problems and knowledge for solving identified problems. This has five-dimensional space namely the physical, digital, data, knowledge and social space. This framework is expected to support self-thinking, self-decision-making, self-execution and self-improving. Cheng et al. (2018) also present the aims of a smart factory for the fourth manufacturing generation. In this case, the digital twin concept is used to achieve physical connection and data collection, virtual models and simulations, data and information technology systems integration and lastly, databased production operations and management methods. These expectations are also embraced by other authors like Ellgass et al. (2018), Qi et al. 2018a) and Zhang et al. (2019a).