Preface
FLINS, an acronym for Fuzzy Logic and Intelligent Technologies in Nuclear Science, is an international research forum intended to advance the theory and applications of computational intelligence for applied research in general and for nuclear engineering in particular. In this framework, the fifth international FLINS conference on Computational Intelligent Systems for Applied Research was successfully held in Gent, Belgium, September 16–18, 2002.
There were more than 70 papers presented at the conference. The topics covered various areas of the trends in computational intelligence in control, decision-making, and nuclear engineering. Especially, Professor L.A. Zadeh (University of California at Berkeley) delivered his plenary lecture on “It is a fundamental limitation to base probability theory on bivalent logic”, Professor R.E. Uhrig (University of Tennessee at Knoxville) outlined in his plenary lecture “Trends in computational intelligence in nuclear engineering”, Professor H. Prade (Institut de Recherche en Informatique de Toulouse) gave his plenary lecture on “Possibility theory in decision support systems”, and Professor P.P. Wang (Duke University at Durham) demonstrated in his plenary lecture on “Computing with words—semantics”. With a rigorous review process, we have selected 14 papers from FLINS 2002 related to the topic of this special issue on information fusion and decision-making under uncertainties.
From a multi-criteria point of view, classical crisp dimension refers to a minimal representation of crisp partial orders as the intersection of linear orders, meaning each one of these linear orders is a possible underlying criterion. The first paper “Crisp dimension theory and valued preference relations” by Gómez et al., analyzes the concept of generalized dimension function of valued preference relations, and presents some properties of such a generalized dimension function. The authors point out that their approach allows alternative representations depending on some underlying rationality core the decision maker may change.
Complex control systems are characterised by probabilistic and fuzzy uncertainty due to which it is hard to determine the correlation between them and their influence on the output. As a result of the difficulties in representing both these basic types of uncertainty, one of them is often ignored in practice. This inadequacy between the nature of the problem and the methods applied has a negative impact on the effectiveness of the models used for these systems. The second paper “Analysis of systems under probabilistic and fuzzy uncertainty using multi-valued logic” by Gegov et al., introduces a special class of functions characterized by two interrelated mappings—logical and probabilistic. This formal theory is implemented by means of input–output sets of data from real multi-factor, multi-stage, and non-stationary processes operating under stochastic and fuzzy uncertainty of the environment. It is shown how incorrect data can be successfully processed for prediction and control purposes by applying interpolation, composition and extrapolation procedures to these functions.
The need for rigorous methodologies for the analysis of the system dependability becomes crucial, particularly with the exponential increase in system complexity and the involvement of these systems in critical scenarios where damage may have drastic consequences on the whole society. The third paper by Oussalah and Newby focuses on a representation of system reliability in the framework of possibility theory. The authors have proposed a new construction of a possibility distribution underlying the time to failure of a given system, given the quantitative (probabilistic) knowledge of system failure as well as some qualitative evidence about the trust on the former knowledge. The study shows the behaviour of the possibilistic mean time to failure, as indicated by the elaborated possibility reliability function, with respect to the (probabilistic) mean time to failure. The induced results reinforce some psychological analyses that stipulate an increase or decrease in probabilistic results when dealing with sufficiently high/low situations. Comparison with results obtained by using probability/possibility transformations shows a difference in allocating the overconfidence (under-confidence) knowledge.
Association rule mining is one of the focal points in research on knowledge discovery. While conventional approaches usually deal with databases with binary values, the fourth paper by Chen et al., presents an approach to discovering association rules from quantitative datasets, which are commonly seen in real-world applications. Fuzzy logic is applied to “discretization” of quantitative domains as well as to logic implication operations so as to remedy possible boundary problems due to sharp partitioning and facilitate fuzzy implication, respectively. Simple rules are discussed and incorporated as optimization strategies into the approach. Experiments with synthetic data as well as with real datasets are carried out to show the performance of the proposed algorithm.
The growing technical complexity of large engineering systems such as offshore platforms and offshore support vessels, together with the intense public concern over their safety has stimulated the research and development of novel safety analysis methods and safety assessment procedures. The fifth paper aims at proposing a framework for modelling safety of engineering systems with various types of uncertainties using a fuzzy rule-based evidential reasoning (FRB-ER) approach. In the framework, parameters used to define the safety level, including failure rate, failure consequence severity and failure consequence probability, are described using fuzzy linguistic variables; a fuzzy rule-base designed on the basis of a belief structure is used to capture uncertainty and nonlinear relationships between these parameters and the safety level; and the inference of the rule-based system is implemented using the evidential reasoning algorithm. A numerical study of collision risk analysis for a kind of offshore platform is used to illustrate the application of the proposed approach.
It is well known that a statistic model is a statistical description of an underlying system, intended to match a real situation as closely as possible. In general, the model for a population is fitted to a sample by estimating the parameters in the model. If the assumption of the model is correct and the size of the sample is large, a statistician can get a perfect statistical model. The larger the size, the more precise the statistical model. However, in many cases, it is very difficult to get a correct assumption and a large sample. Hence many of the non-parametric models and the interval estimation have been suggested. The sixth paper by Huang demonstrates the reliability of interior-outer-set mode by executing some computer simulation experiments. The simulation results show that the interior-outer-set model is more reasonable than the histogram model because the error from the interior-outer set model is smaller. The most important issue is that the possibility-probability distribution (calculated by using the interior-outer-set model) coordinates the practical situation: it is impossible to assign a unique probability value for an event with incomplete information.
The next paper on “An approach to calculate optimal window-width serving for information diffusion technique” by Wang et al., presents an approach to obtain the optimal window-width that can be developed into a powerful tool to promote the study of the information the diffusion technique.
There is a need for methods that will extract accurate results from limited data in information processing. The eighth paper on “Unlimited information diffusion method and application in risk analysis in coronary heart disease” by Shang et al., probes into the mechanism of unlimited information diffusion, which is similar to those of molecular diffusion and heat condition. Therefore the information diffusion function is deduced to be the solution of the Cauchy problem. The authors apply this information diffusion function to study the relationship between the prevalence rates of coronary heart disease and the relevant risk factors in a practical project. The results are satisfactory and reveal that the information diffusion technique is efficient in dealing with the small sample problem. To get rid of the randomicity of the parameter in the information diffusion function, two criteria are proposed for the establishment of an optimization model in the one-dimensional case.
In the textile/garment industry, fabric hand evaluation is generally performed in two ways: (1) subjective evaluation and (2) objective evaluation. The subjective evaluation is the subjective feeling when experts touch a fabric with their hands. It is based on human sensitivity and experience and it varies with time, personal preference and cultural background. It is strongly related to consumer preference but difficult to be quantized. So, this evaluation restricts the scientific understanding of fabric performance for those who wish to design high-quality fabrics by engineering means. The objective evaluation is performed by measuring a set of physical features on fabrics. The objective evaluation leads to precise numerical data describing indirectly fabric hand but its interpretation with respect to human feeling should be further exploited. The ninth paper by Zeng et al., presents a new method for modeling the relationship between objective and subjective fabric hand evaluation and between adjusting parameters of fabric production and objective fabric hand features. In this method, fuzzy techniques, principal component analysis and other data analysis techniques are applied and human knowledge on fabric production and numerical data measured on instruments are used in a complementary way for selecting relevant physical features and extracting fuzzy rules. The effectiveness of this method has been shown through a number of knitted cotton samples.
Today's markets are generally perceived to be demanding much higher-quality and higher-performing products at a reasonably low cost. Software development is a special case for product development. The tenth paper by Büyüközkan et al., proposes a methodology to improve the quality of decision-making in the software development project under uncertain conditions. To deal with the uncertainty and vagueness from subjective perception and experience of humans in the decision process, a methodology based on the extent fuzzy analytic hierarchy process modeling to assess the adequate economic (tangible) and quality (intangible) balance is applied. Two key factors of economic and quality are evaluated separately by fuzzy approaches and both factors' estimates are combined to obtain the preference degree associated with each software development project strategy alternatives for selecting the most appropriate one. Using the proposed approach, the ambiguities involved in the assessment data can be effectively represented and processed to assure a more convincing and effective decision-making.
In the eleventh paper, de Moraes and Santos Machado present a new approach to online evaluation in virtual reality simulators. This approach has an elegant mathematical formalism of continuous hidden Markov models and fuzzy sets for modeling and classification of the trainee in pre-defined classes of training.
Melin and Castillo describe in the twelfth paper adaptive model-based control of non-linear plants using type-2 fuzzy logic and neural networks. A specific non-linear plant was used to simulate the hybrid approach for adaptive control and was used as test bed in the experiments of the two-link robot arm. The results of the type-2 fuzzy logic approach for control were good, both in accuracy and efficiency.
By the same authors, Castillo and Melin in the next paper present a new approach for plant monitoring and diagnostics using type-2 fuzzy logic and fractal theory. The authors also compare the results of the type-2 fuzzy logic approach with those of using only a traditional type-1 fuzzy logic approach. Experimental results show a significant improvement in the monitoring ability with the type-2 fuzzy logic approach.
Knowledge-based control tries to integrate the knowledge of human operators or process engineers into the controller design. Fuzzy control, one of the most popular intelligent techniques, has been successfully applied to a large number of consumer products and industrial processes. Model predictive control (MPC) has been used in process control systems with constraints; however, the constrained optimization problem involved in control systems has generally been solved in practice in a piece-meal fashion. To solve the problem systemically, the Multi-Objective Fuzzy-Optimization (MOFO) is used in the constrained predictive control for online applications as a means of dealing with fuzzy goals and fuzzy constraints in control systems. The conventional model predictive control is integrated with the techniques from fuzzy multicriteria decision-making, translating the goals and the constraints to predictive control in a transparent way. The last paper by Li et al., investigates how to use the fuzzy goal programming in predictive control and how to use the fuzzy goals and fuzzy constraints in predictive control. The presented method allows for a more flexible aggregation of the control objectives than the usual weighting sum of squared errors. It is shown that the model predictive controller based on MOFO allows the designers a more flexible aggregation of the control objectives than the usual weighting sum of squared errors in MPC. The visual robot path planning validates the efficiency of the presented algorithm.
This special issue is an ongoing progress report of FLINS activities and especially the result of FLINS 2002. Special thanks are due to all contributors and all referees for their kind co-operation in helping to prepare this volume, and to Professor George Klir (Editor-in-Chief) and Monika Fridrich (Editorial Assistant) of International Journal of General Systems for their consideration, help, and advice to publish this special issue of their journal.
Da Ruan
Guest Editor
Mol, Belgium
August 2003