Neural Networks Analysis of Steel Plate Processing Augmented by Multi‐objective Genetic Algorithms

An earlier neural network analysis of processing of steel plates through hot rolling was subjected to a further refined analysis through some flexible neural networks that evolved using a multi‐objective predator‐prey genetic algorithm. The original data set expressing the Yield Strength and Ultimate Tensile Strength of the rolled slabs in terms of a total of 108 process variables were subjected to a systematic pruning through this evolutionary approach, till the nitrogen content of the steel emerged as the most significant input variable. A theoretical explanation is provided for this slightly unexpected result.     

Decision making with a fuzzy ontology

Knowledge mobilisation is a transition from the prevailing knowledge management technology that has been widely used in industry for the last 20?years to a new methodology and some innovative methods for knowledge representation, formation and development and for knowledge retrieval and distribution. Knowledge mobilisation aims at coming to terms with some of the problems of knowledge management and at the same time to introduce new theory, new methods and new technology. More precisely, this paper presents an outline of a fuzzy ontology as an enhanced version of classical ontology and demonstrates some advantages for practical decision making. We show that a number of soft computing techniques, e.g. aggregation functions and interval valued fuzzy numbers, will support effective and practical decision making on the basis of the fuzzy ontology. We demonstrate the knowledge mobilisation methods with the construction of a support system for finding the best available wine for a number of wine drinking occasions using a fuzzy wine ontology and fuzzy reasoning methods; the support system has been implemented for a Nokia N900 smart phone.     

Furfural Oxidation with Hydrogen Peroxide Over ZSM-5 Based Micro-Mesoporous Aluminosilicates

Micro-mesoporous aluminosilicates based on ZSM-5 zeolite, obtained by a dual template method, as well as in the presence of a dual-functional template (i.e. a Gemini-type surfactant), were tested in the oxidation of furfural with hydrogen peroxide. Even substantial changes in acidity and porosity of the catalysts result in minor variations of selectivity towards the desired products. Application of the synthesized zeolite-based materials in the oxidation of furfural with hydrogen peroxide leads to formation of 2(5H)-furanone (yield up to 28.5%) and succinic acid (up to 19.5%) as the main C4 reaction products. The kinetic model developed previously to treat the results for oxidation of furfural over sulfated zirconia was able to describe the data also for micro-mesoporous aluminosilicates.

Graphical Abstract

     

Towards a neighborhood simplification of tile systems: From Moore to quasi-linear dependencies

Self-assembly is the process by which objects aggregate independently and form complex structures. One of the theoretical frameworks in which the process of self-assembly can be embedded and formally studied is that of tile systems. A Wang tile is a square unit, with glues on its edges, attaching to other tiles which have matching glues, and forming larger and larger structures. In this paper we concentrate over two basic, but essential, self-assembling structures done by Wang tiles. The first one, called ribbon, is a non-self-crossing wire-like structure, in which successive tiles are adjacent along an edge, and where tiles are glued to their predecessor and successor by use of matching glues. The second one, called zipper, is a similar contiguous structure, only that here, all touching tiles must have matching glues on their abutting edges, independently of their position in the structure. In case of Wang tiles, it has been shown that these two structures are equivalent. Here we generalize this result for the case when the tiles have eight glues, four on their edges and four on their corners. Thus we show that an eight neighborhood dependency, namely the Moore neighborhood, can be simulated by a quasi-linear dependency.     

Geno-mathematical identification of the multi-layer perceptron

In this paper, we will focus on the use of the three-layer backpropagation network in vector-valued time series estimation problems. The neural network provides a framework for noncomplex calculations to solve the estimation problem, yet the search for optimal or even feasible neural networks for stochastic processes is both time consuming and uncertain. The backpropagation algorithm—written in strict ANSI C—has been implemented as a standalone support library for the genetic hybrid algorithm (GHA) running on any sequential or parallel main frame computer. In order to cope with ill-conditioned time series problems, we extended the original backpropagation algorithm to a K nearest neighbors algorithm (K-NARX), where the number K is determined genetically along with a set of key parameters. In the K-NARX algorithm, the terminal solution at instant t can be used as a starting point for the next t, which tends to stabilize the optimization process when dealing with autocorrelated time series vectors. This possibility has proved to be especially useful in difficult time series problems. Following the prevailing research directions, we use a genetic algorithm to determine optimal parameterizations for the network, including the lag structure for the nonlinear vector time series system, the net structure with one or two hidden layers and the corresponding number of nodes, type of activation function (currently the standard logistic sigmoid, a bipolar transformation, the hyperbolic tangent, an exponential function and the sine function), the type of minimization algorithm, the number K of nearest neighbors in the K-NARX procedure, the initial value of the Levenberg–Marquardt damping parameter and the value of the neural learning (stabilization) coefficient α. We have focused on a flexible structure allowing addition of, e.g., new minimization algorithms and activation functions in the future. We demonstrate the power of the genetically trimmed K-NARX algorithm on a representative data set.     

Data Refinement of Remote Procedures

Recently the action systems formalism for parallel and distributed systems has been extended with the procedure mechanism. This gives us a very general framework for describing different communication paradigms for action systems, e.g. remote procedure calls. Action systems come with a design methodology based on the refinement calculus. Data refinement is a powerful technique for refining action systems. In this paper we will develop a theory and proof rules for the refinement of action systems that communicate via remote procedures based on the data refinement approach. The proof rules we develop are compositional so that modular refinement of action systems is supported. As an example we will especially study the atomicity refinement of actions. This is an important refinement strategy, as it potentially increases the degree of parallelism in an action system. Received February 1999 / Accepted in revised form July 2000     

Invariant diagrams with data refinement

Invariant based programming is an approach where we start to construct a program by first identifying the basic situations (pre- and post-conditions as well as invariants) that could arise during the execution of the algorithm. These situations are identified before any code is written. After that, we identify the transitions between the situations, which will give us the flow of control in the program. Data refinement is a technique of building correct programs working on concrete data structures as refinements of more abstract programs working on abstract data types. We study in this paper data refinement for invariant based programs and we apply it to the construction of the classical Deutsch–Schorr–Waite graph marking algorithm. Our results are formalized and mechanically proved in the Isabelle/HOL theorem prover.     

Characterisation of CO/NO/SO2 emission and ash-forming elements from the combustion and pyrolysis process

Bioenergy is considered as a sustainable energy which can play a significant role in the future’s energy scenarios to replace fossil fuels, not only in the heat production, but also in the electricity and transportation sectors. Emission formation and release of main ash-forming elements during thermal conversion of biomass fuels at different conditions have been the scope of this study. The experiments were conducted in a quartz glass reactor where the temperature and atmosphere could be controlled. The selected fuels represent a wide range of biomass compositions. They are torrefied softwood, spruce bark, waste wood, miscanthus, and wheat straw. The fuels were first grinded and then pressed with a pellet maker into pellets of the same size and weight. For each fuel, the experiments were carried out under both oxidation and pyrolysis condition, with atmosphere of 3 % O₂ + 97 % N₂ and 100 % N₂, respectively, at four residence times. The selected temperatures under which experiments were performed are 800, 900, and 1,050 °C. The concentration of SO₂, NO, CO, and CO₂ emissions and O₂ were monitored online by three analysers, simultaneously. The residue weight was measured after each process, and the comparison with the ash content of the fresh pellet is made. Additionally, the release of several ash-forming elements (K, Zn, Na, and Mn) from the fuels has been quantified as function of temperature and residence time by inductively coupled plasma mass spectrometry (ICP-MS) and inductively coupled plasma atomic emission spectroscopy (ICP-AES). Time-dependent formation of NO and SO₂ and other emissions is presented and discussed with respect to different temperature and combustion conditions.     

State space modelling and spectral analysis of cointegrated vector processes (evidence from the U.S. and Scandinavian economies)

We provide new evidence on the impact of the U.S. economy on two Scandinavian economies (Finland and Sweden). Initially, we test for the presence of unit roots among the observed input-output processes. Next, Granger causality and cointegra-tion of the system is explicitly tested, to justify the estimated vector-valued state space model. The trend and cyclical components of the endogenous vector are extracted by three alternative decomposition methods. Finally, the content of the cyclical component is analysed by spectral analysis.