In current neural network research, simulation plays a crucial role. Although there is a wide range of neural network simulators available, it is impossible to keep up with the continuous surge of new neural networks and their variations. Consequently, the extensibility and modularity of neural network simulation software is an important issue. Implementation and modification of neural networks and their embedding into an simulation environment should be possible with minimal effort.

In manufacturing, the determination of a (near) optimal sequence of machining operations to create a part is a non-trivial task. The paper presents the use of datum hierarchy trees in operation sequencing to ensure the finished part satisfies the design tolerances. Based on this technique, a framework is proposed to automate not only the retrieval of relevant plans for evolving the datum hierarchy tree of a new part, but also the optimisation of the operation sequence subjected to constraints on group cell layout, cut sequence, and machine tolerances. The retrieval process is based on the classification of parts using a back propagation neural network while the operation sequence is optimised using an evolutionary algorithm. Results showed that both geometrical and part features may be needed to assist the categorisation of the parts investigated. The results from a case study using industry parts from an aerospace company revealed the potential practical value of the proposed approach in deriving operation sequence for minimal machine and datum changes, both including and excluding manufacturing sequence constraints imposed by group cell layout.

The focus of this publication is a review of the state of the art in tolerance analysis, synthesis, and transfer for geometric and dimensional tolerances in sheet metal forming and the integration solutions with computer-aided process planning systems. In this context, the general tolerance methods are first described. Then, the mathematical models for sheet metal tolerance analysis and synthesis are examined in detail. To address the CAPP modeling concerns, the paper is then followed up with a brief review of past research works related to feature-based process planning. Finally, those imperative future research areas are identified.

The link between the crystal topology and symmetry is examined, focusing on the conditions under which a structure with a given topology can exhibit a certain symmetry. By defining embeddings for quotient graphs (finite representations of crystal topologies) and the corresponding nets (the graph-theoretical equivalents of structures), a strong relationship between the automorphisms of the quotient graphs and the symmetry of the embedded net is established. This allows one to constrain the relative node positions under the premise that an embedding of a net has a certain symmetry, and allows one to assign nodes to equivalents of Wyckoff positions. Two-dimensional examples as well as known crystal structures are used to illustrate the findings. A comparison with a related publication and a discussion on whether constraints on distances between atoms and on bond angles result in restrictions on symmetry without causing confusion conclude the work.

Product lifecycle stages are inter-related and mutually constraining. Due to the sequential nature of the product development processes, some constraints or conflicts may emerge in a later stage and require modifications to the decisions made in earlier stages. The iterations between stages are hence unavoidable and must be managed carefully to maintain the consistency, integrity and validity of product information models. Due to the inter- or intra-stage relations, a chain of changes is very likely to occur as the consequence of an initial change. Modeling and maintaining these relations are important in collaborative engineering to evolve the state of the whole product model in a consistent manner. This paper introduces a new method of modeling associative engineering relations in a unified feature modeling scheme and elaborates a change propagation algorithm for the information consistency control among multiple applications of product lifecycle stages. The algorithm is established on a JTMS-based dependency network. Two case studies are used to illustrate the proposed dependency network and change propagation algorithm.

With widely used concurrent and collaborative engineering technologies, the validity and consistency of product information become important. In order to establish the state of the art, this paper reviews emerging concurrent and collaborative engineering approaches and emphasizes on the integration of different application systems across product life cycle management (PLM) stages. It is revealed that checking product information validity is difficult for the current computer-aided systems because engineering intent is at best partially represented in product models. It is also not easy to maintain the consistency among related product models because information associations are not established. The purpose of this review is to identify and analyze research issues with respect to information integration and sharing for future concurrent and collaborative engineering. A new paradigm of research from the angle of feature unification and association for product modeling and manufacturing is subsequently proposed.

This publication discusses the integration of the conceptual design and the end-of-life stages in the management of a product. This is done by pointing out the influences of the end-of-life of a product on the cost of a product and investigating links between these two stages. Then, both stages and their interrelations are examined for possible needs in terms of their management. This knowledge is then used to evaluate the status quo of product life-cycle management. In particular, the gap between the "as-is" and "should be" functionality of PLM solutions with respect to these two stages is examined for two most popular PLM solutions.

A literature review of sheet metal forming errors as well as geometrical dimensions and tolerances (GD&T) shows that the theoretical means for the allocation of process tolerances with respect to GD&T are insufficient. In order to judge the influence of geometrical process errors (e.g., angular errors of bends), two typical sheet metal designs with parallelism and a position tolerance are studied. These case studies comprise a detailed analysis of tolerance chains including angular errors of bends and their positions. The resulting errors are compared with those resulting from length dimensional process errors and conclusions are drawn.

Features allow one to associate human knowledge and product geometry. The authors proposed, in earlier publications, a unified feature modeling scheme with the aim to maintain the integrity and consistency of a product model. Different application feature models within and across different product life-cycle stages are integrated, and especially, nongeometric relations (besides geometric ones) are handled. In this paper, as an improvement to the previous work, two types of associations are introduced: sharing and dependency. In the context of conceptual and detail design stages, these associations are described and the implementation is discussed in detail.

Quotient graphs and nets — the graph theore tical correspondences of cells and crystal structures — are reintroduced independent from crystal structures. Based on this, the issue of iso- and automorphism of nets, the graph theoretical equivalent of symmetry operations, is closely examined. A result, it is shown that the topology of a net (that is the bonds in a crystal) constrains severely the symmetry of the embedding (that is the crystal), and in the case of connected nets the space group except for the setting. Several examples are studied and conclusions on phases are drawn (pseudo-cubic FeS2 versus pyrite; α-versus β-quartz; marcasite-versus rutile-like phases).

This paper presents the potential of modelling a product's life-cycle using the Unified Modelling Language (UML). The potential benefits and limitations are discussed. An example of a vacuum cleaner is cited in support of this approach. Model consistency across the various life cycle stages of the product is of major concern and an algorithm for constraint management is proposed and prospective research directions highlighted.

AbstractDifferent computer-aided systems, which are used in specific product lifecycle stages, have different requirements on product geometry representation. Non-manifold and multi-dimensional geometries are also needed besides two-manifold solid models. Traditional geometric modeling systems, which usually use B-rep or CSG solid models, have limitations to accommodate these diversified requirements. In addition, in an integrated product development environment, to share data as well as to propagate changes across applications, a unified feature modeling scheme, which can support different geometric modeling requirements uniformly, is preferred. In this paper, a unified, cellular topology based feature modeling scheme is proposed. Its model structure and usage in integrating application models are described.

Computer-Aided Design and Applications is an international journal on the applications of CAD and CAM. It publishes papers in the general domain of CAD plus in emerging fields like bio-CAD, nano-CAD, soft-CAD, garment-CAD, PLM, PDM, CAD data mining, CAD and the internet, CAD education, genetic algorithms and CAD engines. The journal is aimed at all developers and users of CAD technology to ptovide CAD solutions for various stages of design and manufacturing. The journal publishes all about Computer-Aided Design and Computer-Aided technologies.

Proper initialization is one of the most important prerequisites for fast convergence of feed-forward neural networks like high order and multilayer perceptrons. This publication aims at determining the optimal value of the initial weight variance (or range), which is the principal parameter of random weight initialization methods for both types of neural networks. An overview of random weight initialization methods for multilayer perceptrons is presented. These methods are extensively tested using eight real-world benchmark data sets and a broad range of initial weight variances by means of more than 30, 000 simulations, in the aim to find the best weight initialization method for multilayer perceptrons. For high order networks, a large number of experiments (more than 200, 000 simulations) was performed, using three weight distributions, three activation functions, several network orders, and the same eight data sets. The results of these experiments are compared to weight initialization techniques for multilayer perceptrons, which leads to the proposal of a suitable weight initialization method for high order perceptrons. The conclusions on the weight initialization methods for both types of networks are justified by sufficiently small confidence intervals of the mean convergence times.

Currently, collaborations in Product Lifecycle Management (PLM) become a major trend. However, different product lifecycle phases are interrelated. Although collaborative product development ap- proach has been suggested to find a balance among usually conflicted requirements of these phases, but very few in-depth research works have addressed the inherent sequential and iterative natures of the product de- velopment processes coherently. Due to the inter- or intra-phase relationships, a chain of changes is very likely to occur. Modeling and maintaining these relations are hence important in collaborative engineering to evolve the state of the whole system in a stable and consistent manner. This paper elaborates a feature-based, object-oriented and unified approach for the information sharing among multiple applications of PLM col- laboration.

Both size and geometric dimension specifications contribute to the manufacturability of a part and must therefore be determined carefully to allow for efficient process plans. The presented approach is meant to systematically explore alternative geometric and size dimensions (as standardized in ANSI Y14.5) with the aim of improving the manufacturability of a design and how well-suited it is for a computer-based application.

A general systematic method of predicting hypothetical crystal structures could enable important advances in many areas of science. We describe a recently developed approach based on graph theory and density functional theory and apply it to enumerate systematically a number of sp(3)-hybridized carbon polymorphs with four atoms per unit cell. The calculations predict three unknown structures that are potentially metastable under appropriate pressure and temperature conditions. The theoretical properties of these hypothetical polymorphs and their relative stability with respect to diamond are discussed.

This paper proposes an agent-based composable simulation framework to address the challenges of integration, composability, distributed coordination, and interaction for the development of a virtual prototype of fluid power system. The approach proposed represents each virtual hydraulic component by a domain agent (DA). The agents are then gathered into a multi-agent system, which models the hydraulic system as a whole. The virtual prototyping evaluation depends on the communication and collaboration of multiple agents. A case study shows that agent-based composable simulation can predict the overall system performance. A prototype implementation of the proposed system is presented in this paper.