George Panoutsos

A neural-fuzzy modelling framework based on granular computing: Concepts and applications

Fuzzy and neural-fuzzy systems have successfully and extensively applied to solve problems in many research areas such as those associated with industrial, medical and academic applications. However, recent trends reveal a demand for a workflow with a particular emphasis on transparency, simplicity, system interpretability as well as on a strong link with human cognition. Such requirement is mainly driven by research areas where expert knowledge is of very high importance and any new proposed modelling system falls under the interpretability scrutiny of experts in order to confirm the systems validity. The relatively recent paradigm of granular computing (GrC) offers an ideal opportunity for a transparent knowledge discovery methodology to be combined with fuzzy logic thereby towards a systematic modelling framework with a focus on the overall transparency of the system. Such transparency in the workflow allows for better interaction between the expert process knowledge and the system design which translates into a better performing system. In this paper a systematic modelling approach using granular computing (GrC) and neural-fuzzy modelling is presented. In this research study a GrC algorithm is used to extract relational information and data characteristics out of an initial database. The extracted knowledge and granular features are then translated into a linguistic rule-base of a fuzzy system. This rule-base is finally elicited and optimised via a neural-fuzzy modelling structure. During the various steps of this methodology the transparency features are highlighted and it is shown here how the system designer can take advantage of such features to enhance the system. The proposed modelling framework is applied to a multi-dimensional and complex data set consisting of measurements of mechanical properties of heat treated steel. The data set is collected from a real industrial process and the measurements are dictated by customer production demands and the data set is very sparse with many discontinuities. The proposed framework successfully models the mechanical properties of heat treated steel and it further improves upon the performance of previously established modelling structures.

Granular computing neural-fuzzy modelling: A neutrosophic approach

Granular computing is a computational paradigm that mimics human cognition in terms of grouping similar information together. Compatibility operators such as cardinality, orientation, density, and multidimensional length act on both in raw data and information granules which are formed from raw data providing a framework for human-like information processing where information granulation is intrinsic. Granular computing, as a computational concept, is not new, however it is only relatively recent when this concept has been formalised computationally via the use of Computational Intelligence methods such as Fuzzy Logic and Rough Sets. Neutrosophy is a unifying field in logics that extents the concept of fuzzy sets into a three-valued logic that uses an indeterminacy value, and it is the basis of neutrosophic logic, neutrosophic probability, neutrosophic statistics and interval valued neutrosophic theory. In this paper we present a new framework for creating Granular Computing Neural-Fuzzy modelling structures via the use of Neutrosophic Logic to address the issue of uncertainty during the data granulation process. The theoretical and computational aspects of the approach are presented and discussed in this paper, as well as a case study using real industrial data. The case study under investigation is the predictive modelling of the Charpy Toughness of heat-treated steel; a process that exhibits very high uncertainty in the measurements due to the thermomechanical complexity of the Charpy test itself. The results show that the proposed approach leads to more meaningful and simpler granular models, with a better generalisation performance as compared to other recent modelling attempts on the same data set.

Interval Type-2 Radial Basis Function Neural Network: A Modeling Framework

In this paper, an interval type-2 radial basis function neural network (IT2-RBF-NN) is proposed as a new modeling framework. We take advantage of the functional equivalence of radial basis function neural networks (RBF-NNs) to a class of type-1 fuzzy logic systems (T1-FLS) to propose a new interval type-2 equivalent system; it is systematically shown that the type equivalence (between RBF and FLS) of the new modeling structure is maintained in the case of the IT2 system. The new IT2-RBF-NN incorporates interval type-2 fuzzy sets within the radial basis function layer of the neural network in order to account for linguistic uncertainty in the systems variables. The antecedent part in each rule in the IT2-RBF-NN is an interval type-2 fuzzy set, and the consequent part is of Mamdani type with interval weights, which are used for the Karnik and Mendel type-reduction process in the output layer of the network. The structural and parametric optimization of the IT2-RBF-NN parameters is carried out by a hybrid approach that is based on estimating the initial rule base and footprint of uncertainty (FOU) directly via a granular computing approach and an adaptive back propagation approach. The effectiveness of the new modeling framework is assessed in two parts. First, the IT2-RBF-NN is tested against a number of popular benchmark datasets, and second, it is demonstrated in a real-world industrial application that has particular challenges that are related to the uncertainty of the raw information. Via simulation results, it is shown that the proposed modeling framework performs well as compared with its T1 equivalent system. In addition, a very good computational efficiency is demonstrated as a result of the systematic and automatic creation of IT2 linguistic information and the FOU. Crucially, the proposed modeling framework opens up a host of opportunities for the academic community that already uses the popular T1-RBF-NN-based structure to try the new IT2-RBF-NN and take advantage of the numerous existing RBF-based adaptive learning algorithms, RBF-based multiobjective optimization techniques, granular computing-based information capture techniques, and real-world FLS implementations, and, in general, take advantage of the computational efficiency of the fusion of IT2-FLS and RBF-NN.

Modeling and optimal design of machining-induced residual stresses in aluminium alloys using a fast hierarchical multiobjective optimization algorithm

The residual stresses induced during shaping and machining play an important role in determining the integrity and durability of metal components. An important issue of producing safety critical components is to find the machining parameters that create compressive surface stresses or to minimize tensile surface stresses. In this article, a systematic data-driven fuzzy modeling methodology is proposed, which allows constructing transparent fuzzy models considering both accuracy and interpretability attributes of fuzzy systems. The new method employs a hierarchical optimization structure to improve the modeling efficiency, where two learning mechanisms cooperate together: the Nondominated Sorting Genetic Algorithm II (NSGA-II) is used to improve the models structure, while the gradient descent method is used to optimize the numerical parameters. This hybrid approach is then successfully applied to the problem that concerns the prediction of machining induced residual stresses in aerospace aluminium alloys. Based on the developed reliable prediction models, NSGA-II is further applied to the multiobjective optimal design of aluminium alloys in a “reverse-engineering” fashion. It is revealed that the optimal machining regimes to minimize the residual stress and the machining cost simultaneously can be successfully located.

Absolute Electrical Impedance Tomography (aEIT) Guided Ventilation Therapy in Critical Care Patients: Simulations and Future Trends

Thoracic electrical impedance tomography (EIT) is a noninvasive, radiation-free monitoring technique whose aim is to reconstruct a cross-sectional image of the internal spatial distribution of conductivity from electrical measurements made by injecting small alternating currents via an electrode array placed on the surface of the thorax. The purpose of this paper is to discuss the fundamentals of EIT and demonstrate the principles of mechanical ventilation, lung recruitment, and EIT imaging on a comprehensive physiological model, which combines a model of respiratory mechanics, a model of the human lung absolute resistivity as a function of air content, and a 2-D finite-element mesh of the thorax to simulate EIT image reconstruction during mechanical ventilation. The overall model gives a good understanding of respiratory physiology and EIT monitoring techniques in mechanically ventilated patients. The model proposed here was able to reproduce consistent images of ventilation distribution in simulated acutely injured and collapsed lung conditions. A new advisory system architecture integrating a previously developed data-driven physiological model for continuous and noninvasive predictions of blood gas parameters with the regional lung function data/information generated from absolute EIT (aEIT) is proposed for monitoring and ventilator therapy management of critical care patients.

GRANULAR COMPUTING AND EVOLUTIONARY FUZZY MODELLING FOR MECHANICAL PROPERTIES OF ALLOY STEELS

Abstract In this paper the development of a model for Mamdani type fuzzy rule-based systems using the new concept of granular computing (GrC) is presented. In this study a GrC algorithm is used to capture the required information in the form of data granules within a high dimensional complex database. The initial collection of information granules is used as a rule-base for a fuzzy inference system (FIS) which is optimised by utilising an Adaptive Genetic Algorithm (AGA). The proposed methodology is applied to real data relating to the heat treatment of alloy steels.

Development of a parsimonious GA-NN ensemble model with a case study for Charpy impact energy prediction

A parsimonious genetic algorithm guided neural network ensemble modelling strategy is presented. Each neural network candidate model to participate in the ensemble model is structurally selected using a genetic algorithm. This provides an effective route to improve the performance of the individual neural network models as compared to more traditional neural network modelling approaches, whereby the neural network structure is selected through some trial-and-error methods or heuristics. The parsimonious neural network ensemble modelling strategy developed in this paper is highly efficient and requires very little extra computation for developing the ensemble model, thus overcoming one of the major known obstacles for developing an ensemble model. The key techniques behind the implementation of the ensemble model, include the formulation of the fitness function, the generation of the qualified neural network candidate models, as well as the specific definitions of the assemble strategies. A case study is presented which exploits a complex industrial data set relating to the Charpy impact energy for heat-treated steels, which was provided by Tata Steel Europe. Modelling results show a significant performance improvement over the previously developed models for the same data set.

Intelligent model-based advisory system for the management of ventilated intensive care patients. Part II: Advisory system design and evaluation

The optimisation of ventilatory support is a crucial issue for the management of respiratory failure in critically ill patients, aiming at improving gas exchange while preventing ventilator-induced dysfunction of the respiratory system. Clinicians often rely on their knowledge/experience and regular observation of the patients response for adjusting the level of respiratory support. Using a similar data-driven decision-making methodology, an adaptive model-based advisory system has been designed for the clinical monitoring and management of mechanically ventilated patients. The hybrid blood gas patient model SOPAVent developed in Part I of this paper and validated against clinical data for a range of patients lung abnormalities is embedded into the advisory system to predict continuously and non-invasively the patients respiratory response to changes in the ventilator settings. The choice of appropriate ventilator settings involves finding a balance among a selection of fundamentally competing therapeutic decisions. The design approach used here is based on a goal-directed multi-objective optimisation strategy to determine the optimal ventilator settings that effectively restore gas exchange and promote improved patients clinical conditions. As an initial step to its clinical validation, the advisory systems closed-loop stability and performance have been assessed in a series of simulations scenarios reconstructed from real ICU patients data. The results show that the designed advisory system can generate good ventilator-setting advice under patient state changes and competing ventilator management targets.

“Right-First-Time” Production: A Reality or a Myth?

There is no doubt that the move towards the development of cost-effective quality products in a highly competitive market has shaped engineering practices and processes in the production of components. The constituents as well as the processing of materials have now become an integral part of the design process where the required mechanical properties have to be “optimized” for a given application and at minimal cost not only to the economy but also to the environment. In the case of steel-making companies, very high standards of surface quality and tighter specifications on hot-rolled strips are constantly being set which, more often than not, result in stringent demands being imposed on the specified metallurgical properties. The quality of the strip product may vary depending on how the strip is processed when it passes through the hot strip mill. It is, therefore, important to develop thermomechanical process models which will allow one to predict the evolution of microstructure and the mechanical properties of the strip during the course of fabrication. The “acid test,” of course, rests with the ability to reverse-engineer (invert) such models in order to control (optimize) the microstructure, and one might argue the ensuing mechanical properties of the material, via the chemistry, the temperature profiles, and the mill schedule. The research activities associated with the Institute for Microstructural and Mechanical Process Engineering: The University of Sheffield (IMMPETUS) which was founded in 1997 lead to truly integrated interdisciplinary research across the disciplines of systems, mechanical, and metallurgical engineering. The article focuses on the integration of systems engineering paradigms for intelligent modeling and optimization of the mechanical properties of materials, in particular steel, to achieve right-first-time production, a challenge for academia in general, and in particular for IMMPETUS.

Adaptive neural-fuzzy inference system for classification of rail quality data with bootstrapping-based over-sampling

An iterative bootstrapping-based data over-sampling strategy is presented in this paper together with an adaptive neural-fuzzy inference system (ANFIS) to deal with a severely imbalanced data modelling problem. As real industrial data are often very large, containing hundreds of process variables and a huge number of data records, the selection of a compact set of input variables becomes critical for any successful modelling and analysis operations. Significant efforts have been devoted to identifying the most relevant input variables through correlation analysis and neural network based forward input selection. An optimal majority to minority class data ratio, which controls the level of data imbalance for model training, is then determined through the iterative bootstrapping process such that the combined sensitivity and specificity performance is optimised. The iterative bootstrapping ANFIS modelling strategy is then applied to a real industrial case study for rail quality classification, with the original data being provided by Tata Steel Europe. Preliminary results show a good overall performance through the iterative bootstrapping data over-sampling ANFIS modelling.