A taxonomy for wavelet neural networks applied to nonlinear modelling

This article presents a novel classification of wavelet neural networks based on the orthogonality/non-orthogonality of neurons and the type of nonlinearity employed. On the basis of this classification different network types are studied and their characteristics illustrated by means of simple one-dimensional nonlinear examples. For multidimensional problems, which are affected by the curse of dimensionality, the idea of spherical wavelet functions is considered. The behaviour of these networks is also studied for modelling of a low-dimension map.

Deflation based nonlinear canonical correlation analysis

This paper introduces two new techniques for determining nonlinear canonical correlation coefficients between two variable sets. A genetic strategy is incorporated to determine these coefficients. Compared to existing methods for nonlinear canonical correlation analysis (NLCCA), the benefits here are that the nonlinear mapping requires fewer parameters to be determined, consequently a more parsimonious NLCCA model can be established which is therefore simpler to interpret. A further contribution of the paper is the investigation of a variety of nonlinear deflation procedures for determining the subsequent nonlinear canonical coefficients. The benefits of the new approaches presented are demonstrated by application to an example from the literature and to recorded data from an industrial melter process. These studies show the advantages of the new NLCCA techniques presented and suggest that a nonlinear deflation procedure should be considered. (c) 2006 Elsevier B.V. All rights reserved.

Learning soft computing control strategies in a modular neural network architecture

Modelling and control of nonlinear dynamical systems is a challenging problem since the dynamics of such systems change over their parameter space. Conventional methodologies for designing nonlinear control laws, such as gain scheduling, are effective because the designer partitions the overall complex control into a number of simpler sub-tasks. This paper describes a new genetic algorithm based method for the design of a modular neural network (MNN) control architecture that learns such partitions of an overall complex control task. Here a chromosome represents both the structure and parameters of an individual neural network in the MNN controller and a hierarchical fuzzy approach is used to select the chromosomes required to accomplish a given control task. This new strategy is applied to the end-point tracking of a single-link flexible manipulator modelled from experimental data. Results show that the MNN controller is simple to design and produces superior performance compared to a single neural network (SNN) controller which is theoretically capable of achieving the desired trajectory. (C) 2003 Elsevier Ltd. All rights reserved.

Robust fuzzy Gustafson-Kessel clustering for nonlinear system identification

This paper deals with Takagi-Sugeno (TS) fuzzy model identification of nonlinear systems using fuzzy clustering. In particular, an extended fuzzy Gustafson-Kessel (EGK) clustering algorithm, using robust competitive agglomeration (RCA), is developed for automatically constructing a TS fuzzy model from system input-output data. The EGK algorithm can automatically determine the 'optimal' number of clusters from the training data set. It is shown that the EGK approach is relatively insensitive to initialization and is less susceptible to local minima, a benefit derived from its agglomerate property. This issue is often overlooked in the current literature on nonlinear identification using conventional fuzzy clustering. Furthermore, the robust statistical concepts underlying the EGK algorithm help to alleviate the difficulty of cluster identification in the construction of a TS fuzzy model from noisy training data. A new hybrid identification strategy is then formulated, which combines the EGK algorithm with a locally weighted, least-squares method for the estimation of local sub-model parameters. The efficacy of this new approach is demonstrated through function approximation examples and also by application to the identification of an automatic voltage regulation (AVR) loop for a simulated 3 kVA laboratory micro-machine system.

A hierarchical physical model-based approach to predictive control of a thermal power plant for efficient plant-wide disturbance rejection

Extending the work presented in Prasad et al. (IEEE Proceedings on Control Theory and Applications, 147, 523-37, 2000), this paper reports a hierarchical nonlinear physical model-based control strategy to account for the problems arising due to complex dynamics of drum level and governor valve, and demonstrates its effectiveness in plant-wide disturbance handling. The strategy incorporates a two-level control structure consisting of lower-level conventional PI regulators and a higher-level nonlinear physical model predictive controller (NPMPC) for mainly set-point manoeuvring. The lower-level PI loops help stabilise the unstable drum-boiler dynamics and allow faster governor valve action for power and grid-frequency regulation. The higher-level NPMPC provides an optimal load demand (or set-point) transition by effective handling of plant-wide interactions and system disturbances. The strategy has been tested in a simulation of a 200-MW oil-fired power plant at Ballylumford in Northern Ireland. A novel approach is devized to test the disturbance rejection capability in severe operating conditions. Low frequency disturbances were created by making random changes in radiation heat flow on the boiler-side, while condenser vacuum was fluctuating in a random fashion on the turbine side. In order to simulate high-frequency disturbances, pulse-type load disturbances were made to strike at instants which are not an integral multiple of the NPMPC sampling period. Impressive results have been obtained during both types of system disturbances and extremely high rates of load changes, right across the operating range, These results compared favourably with those from a conventional state-space generalized predictive control (GPC) method designed under similar conditions.

Constructing networks of continuous-time velocity-based models

A conventional local model (LM) network consists of a set of affine local models blended together using appropriate weighting functions. Such networks have poor interpretability since the dynamics of the blended network are only weakly related to the underlying local models. In contrast, velocity-based LM networks employ strictly linear local models to provide a transparent framework for nonlinear modelling in which the global dynamics are a simple linear combination of the local model dynamics. A novel approach for constructing continuous-time velocity-based networks from plant data is presented. Key issues including continuous-time parameter estimation, correct realisation of the velocity-based local models and avoidance of the input derivative are all addressed. Application results are reported for the highly nonlinear simulated continuous stirred tank reactor process.

Application of static charge transfer within an ionic-liquid force field and its effect on structure and dynamics

The effects of linear scaling of the atomic charges of a reference potential on the structure, dynamics, and energetics of the ionic liquid 1,3-dimethylimidazolium chloride are investigated. Diffusion coefficients that span over four orders of magnitude are observed between the original model and a scaled model in which the ionic charges are +/- 0.5 e. While the three-dimensional structure of the liquid is less affected, the partial radial distribution functions change markedly-with the positive result that for ionic charges of +/- 0.7 e, an excellent agreement is observed with ab initio molecular dynamics data. Cohesive energy densities calculated from these partial-charge models are also in better agreement with those calculated from the ab initio data. We postulate that ionic-liquid models in which the ionic charges are assumed to be +/- 1 e overestimate the intermolecular attractions between ions, which results in overstructuring, slow dynamics, and increased cohesive energy densities. The use of scaled-charge sets may be of benefit in the simulation of these systems-especially when looking at properties beyond liquid structure-thus providing on alternative to computationally expensive polarisable force fields.

Liquid structure of the ionic liquid, 1-methyl-4-cyanopyridinium bis{(trifluoromethyl)sulfonyl}imide determined from neutron scattering and molecular dynamics simulations

The liquid structure of 1-methyl-4-cyanopyridinium bis {(trifluoromethyl)sulfonyl}imide, a prototypical ionic liquid containing an electron-withdrawing group on the cation, has been investigated at 368 K. Experimental neutron scattering combined with empirical potential structure refinement analysis of the data and classical molecular dynamics simulations have been used to probe the liquid structure in detail. Both techniques generated highly consistent results that provide valuable validation of the force fields and refinement approaches. A significant degree of apparent charge ordering is found in the liquid structure, although the nonspherical shape of the ions results in interpenetration of cations into the first shell of adjacent cations, with much shorter closest contact distances than the averaged center-of-mass cation-cation and cation-anion separations.

Thermodynamics, structure, and dynamics in room temperature ionic liquids: The case of 1-butyl-3-methyl imidazolium hexafluorophosphate ([bmim][PF6])

A detailed investigation of the phase diagram of 1-butyl-3-methyl imidazolium hexafluorophosphate ([bmim][PF6]) is presented on the basis of a wide set of experimental data accessing thermodynamic, structural, and dynamical properties of this important room temperature ionic liquid (RTIL). The combination of quasi adiabatic, continuous calorimetry, wide angle neutron and X-ray diffraction, and quasi elastic neutron scattering allows the exploration of many novel features of this material. Thermodynamic and microscopic structural information is derived on both glassy and crystalline states and compared with results that recently appeared in the literature allowing direct information to be obtained on the existence of two crystalline phases that were not previously characterized and confirming the view that RTILs show a substantial degree of order (even in their amorphous states), which resembles the crystalline order. We highlight a strong connection between structure and dynamics, showing the existence of three temperature ranges in the glassy state across which both the spatial correlation and the dynamics change. The complex crystalline polymorphism in [bmim][PF6] also is investigated; we compare our findings with the corresponding findings for similar RTILs. These results provide a strong experimental basis for the exploration of the features of the phase diagram of RTILs and for the further study of longer alkyl chain salts.