Showing papers in "Engineering Applications of Artificial Intelligence in 1996"

TL;DR: An enhanced rough-sets method for generating prediction rules from a set of observed data is presented, extending upon the standard rough set model by making use of the statistical information inherent in the data to handle incomplete and ambiguous training samples.

TL;DR: An efficient hybrid algorithm named ASAGA (Adaptive Simulated Annealing Genetic Algorithm) is proposed, by introducing a mutation operator like simulated annealing and an adaptive cooling schedule to produce an adaptive algorithm that has the merits of both genetic algorithms and simulatedAnnealing.

TL;DR: The architecture of the helicopter fuzzy logic controller is described, the details of the genetic algorithm application are provided, and the results of an actual flight test using the computer software are presented.

TL;DR: It is argued that a knowledge-based approach to FMEA can alleviate most of these problems and the solution chosen in the WIFA project is presented, which employs various knowledge bases to support complete and precise descriptions of processes and products.

TL;DR: A method of computational synthesis of the conceptual design of mechanisms is reported, which employs best-first heuristic searches in a library of mechanical devices, represented and classified qualitatively from various perspectives.

TL;DR: The concept of parallel distributed processing base learning in artificial neural networks is presented, assisting with experimental evidence to predict moment-curvature parameters that are usually accomplished solely by experimental work.

TL;DR: Issues relating to learning stability, training laws and parametric convergence, network conditioning, gradient noise, the curse of dimensionality associated with associative memory networks, automatic network construction algorithms, and a series of neurofuzzy control design laws are discussed.

TL;DR: Methods using artificial neural networks (NN) have been developed for robot inverse compensation, for both local and global calibration problems, and results are presented to show the effectiveness of the NN-based approach.

TL;DR: A framework of a knowledge-based evaluation system for new product concepts, called EIMPPLAN-1, which can select the appropriate plastic material and generate the major injection mold design features, is proposed in the paper.

TL;DR: The main goal of this paper is to present the general framework of a case-based design system for supporting a designer through the earliest steps of designing chemical processes and their representation in flowsheets.

TL;DR: The paper shows that simultaneous optimisation of yarn qualities is easily achieved as a function of the necessary (optimal) input parameters, and that the results are considerably better than current manual machine intervention.

TL;DR: An intelligent fixture design system based on case-based reasoning (CBR) that retrieves the most similar case from the antecedent cases when it faces a new problem, and modifies the case to satisfy the new situation.

TL;DR: A neural network is proposed, based on a functional link net, with two processing modules, designed to implement the calculation of the correlations in functional terms, to classify solder joints on printed-circuit boards (PCB) using a neural-network approach.

TL;DR: An architecture and test results for a computer-based system for assisting the conceptual phase of building design are presented, illustrating how interpretation of the geometric model produces a symbolic model and supports multiple and changing analyses and evaluations during design.

TL;DR: This paper describes research in developing an algorithm for a constraint-satisfaction problem (CSP) which solves certain classes of scheduling problems by formulating them as multi-dimensional placement problems by combining simulation techniques with constraint-based reasoning to handle the stochastic nature of the problem.

TL;DR: Two types of reaction strategies are worked out, based on human experience, and reaction rules governing the system behavior are synthesized corresponding to the different situations defined by the obstacle position, the target orientation and the robot's direction of movement.

TL;DR: The author found that including knowledge of the constraint space into the adaptive searches significantly improved the efficiency of the search process over the simple, untailored application of the methods.

TL;DR: A new model for nonlinear systems which consists of a linear part and a static nonlinear output part is presented, which can be applied to represent a relatively large class of nonlinear dynamic systems with fading memory.

TL;DR: The paper will analyse the problem domain, discuss the need for a computing solution, compare two different computing approaches, and state the effects and potential benefits to industry of introducing such systems.

TL;DR: By adding a harmony model to the LVQ classifier, the proposed method can construct an input-output mapping based on human knowledge and stipulated input—output vector pairs that is useful in classifying input features inherent with overlapping distributions and high uncertainty in the class boundaries.

TL;DR: The method combines elements from variational geometry, matrix algebra and graph theory to construct a composite network describing the structural and topological relations among the various entities that combine the 3D object.

TL;DR: In this article, a hybrid neural network with different activation functions for different layers in fully connected feed-forward neural networks is introduced, where the parameters are the dynamic range, symmetry and slope of the function respectively.

TL;DR: It is shown that using the proposed problem formulation, a non-linear state feedback can also be implemented, which expands the search space for the design, and is computed to show the efficacy of such a method.

TL;DR: Critical issues are identified: the size of the cases, alternatives for problem decomposition, adaptation techniques, solution composition, and overall strategies in case- adaptation systems for synthesis tasks.

TL;DR: The possibility of extending this method to cover the circuitry of a complete car at once is addressed, as are the implications of having not just one such tool, but a toolbox of automated design analysis tools.

TL;DR: This paper describes the experiences of the development team in validating the performance of a large commercial diagnostic knowledge-based system and covers the procedure employed to carry out this task, as well as the heuristic technique used for generating the representative set of test cases.

TL;DR: The authors of this paper have been exploring the hypothesis that teleologicall causal models of engineering designs give rise to the indexing vocabulary, enable and constrain the learning of indices for new designs, and provide similarity measures for matching a target problem with the stored design and retrieving relevant ones.

TL;DR: This paper demonstrates how to apply neural networks with fuzzy logic for document retrieval and uses the Fuzzy Kohonen Neural Network (FKNN) as an example to illustrate the versatility of fuzzy neural networks as applied to the document-retrieval process.

TL;DR: This paper deals with the internal structural design of the PEs at the lower layers of the KYDON architecture, and provides a description of the low-level image-processing tasks performed by KYDON's lower-array processors.

TL;DR: Commercial expert-system shells have the inference engine resident within the shell, and have their knowledge-representation formalisms restricted, so a knowledge-based front end (KBFE) has been developed using LISP to circumvent this restriction.