Selection of network and learning parameters for an adaptive neural robotic control scheme

This paper presents the results of a study in the design of a neural network based adaptive robotic control scheme. The neural network used here is a two hidden layer feedforward network and the learning scheme is the well-known backpropagation algorithm. The neural network essentially provides the inverse of the plant and acts in conjunction with a standard PD controller in the feedback loop. The objective of the controller is to accurately control the end position of a single link manipulator in the presence of large payload variations, variations in the link length and also variations in the damping constant. Based on results of this study, guidelines are presented in selecting the number of neurons in the hidden layers and also the parameters for the learning scheme used for training the network. Results also indicate that increasing the number of neurons in the hidden layer will improve the convergence speed of learning scheme up to a certain limit beyond which the addition of neurons will cause oscillations and instability. Guidelines for selecting the proper learning rate, momentum and fast backpropagation constant that ensure stability and convergence are presented. Also, a relationship between the r.m.s. error and the number of iterations used in training the neural network is established.     

Structural instability of uniaxially compressed α-quartz

We report a molecular dynamics simulation study of stress–strain response of single crystal α-quartz under uniaxial compression using the interatomic potential of van Beest et al. [Phys. Rev. Lett. 64 (1990) 1955]. The results at 1073 K shows an abrupt onset of large deformation of 14 GPa and subsequent time-dependent evolution of an “amorphous” structure with near-planar character. We regard this to be a lattice instability leading to structural failure, to be distinguished from a similar instability observed under hydrostatic loading which results in a phase change. The simulation results allow us to suggest new interpretations of experimental observations with regard to the formation of thin zones of isotropic (glassy) material along faults and a possible mechanism for planar fracture.     

Automated Modeling of Three-Dimensional Structural Components Using Irregular Lattices

Abstract:   Irregular lattices are used to model three-dimensional (3D) structural components consisting of a bulk material, curvilinear reinforcement, and their interface. Domain discretization is highly automated and involves the semi-random placement of nodal points within the domain, followed by Voronoi tessellation of the nodal point set. A technique is given for the Voronoi partitioning of nonconvex domains. For discretizing nonconvex domains, and for effective gradation of nodal point density, a minimum allowable distance between nodes is maintained and the domain is saturated with nodes. To accelerate this computationally expensive operation, a partitioned domain search is used during the filling process. Reinforcement, and its interface with the bulk material, are discretely modeled and freely positioned in the domain, irrespective of the geometry of the irregular lattice representing the bulk material. This method of discretization facilitates model construction, results interpretation, and possible revisions to the model. While the focus is on automated domain discretization and the modeling of reinforcement, elastic properties of the model are demonstrated through examples involving nodal stress calculations and deflection analyses of prestressed concrete beams.     

Scanning Electron Microscopy of Fe79.5B6.5C14 Network Alloys: Part I

Molten Fe_79.5B_6.5C₁₄ melts can be cast into ingots of networklike microstructure. According to microstructural classification, a Fe_79.5B_6.5C₁₄ network specimen can be divided into three zones. Zone A is occupied by a random network, zone B by a uniformly aligned network, and zone C by a severely aligned network that exhibits network patterns.     

A re-examination of seasonal variation in suicides in Australia and New Zealand

BACKGROUND: To examine the seasonality of suicides in Australia and New Zealand during the period 1981 to 1993. METHODS: A chi-square test and a harmonic analysis were used to detect the seasonality of the suicide data. RESULTS: The reduced amplitude and a smaller proportion of variance accounted for by seasonality suggested the seasonal effect on suicide is greatly diminished. The absence of biseasonal distribution of female suicides was also consistently found in the two countries. The finding was contrary to the reported results in seventies in many Western countries. CONCLUSIONS: The change in living condition, roles of males and females and communication pattern resulted in the reduction of climatic and environment effect in the seasonality of suicides were suggested. LIMITATIONS: The results would be better if a longer series of suicide date were available.     

NOTCH EFFECT ON TWO-LEVEL CUMULATIVE LOW-CYCLE FATIGUE LIFE UNDER DIFFERENT BIAXIAL LOADING MODE SEQUENCES

Abstract Two-level cumulative low-cycle fatigue lives of AISI 316 stainless steel notched specimens with different biaxial loading mode sequences were experimentally analyzed in this paper. Forty-eight cases were conducted in the experimental program by considering the loading level sequence effect, the biaxiality of two levels and the cycle ratios. Results show that interlock effect caused by the characteristic fracture surfaces of the different biaxial states is beneficial to the cumulative fatigue lives. On the other hand, the tensile loading of the second level will accelerate the opening of cracks and decrease fatigue strength. Miner's rule predicts most fatigue lives within 30% error bands, and the loading level sequence effect was not found in this research involving complex fracture modes. Fractography of specimens in the cumulative fatigue tests is reported and discussed in this paper.     

Synergistic science

Should computational materials science be recognized as a field with a role in the community comparable to computational physics or chemistry? With the emergence of multiscale modelling, the answer is a resounding 'yes'.