A novel continuous forward algorithm for RBF neural modelling

A continuous forward algorithm (CFA) is proposed for nonlinear modelling and identification using radial basis function (RBF) neural networks. The problem considered here is simultaneous network construction and parameter optimization, well-known to be a mixed integer hard one. The proposed algorithm performs these two tasks within an integrated analytic framework, and offers two important advantages. First, the model performance can be significantly improved through continuous parameter optimization. Secondly, the neural representation can be built without generating and storing all candidate regressors, leading to significantly reduced memory usage and computational complexity. Computational complexity analysis and simulation results confirm the effectiveness.

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.

Sampled-data gain scheduling of continuous LTV plants

This paper discusses the design of gain- scheduled sampled-data controllers for continuous-time polytopic linear parameter-varying systems. The scheduling variables are assumed to available only at the sampling instants, and a bound on the time-variation of the scheduling parameters is also assumed to be known. The resultant gain-scheduled controllers improve the maximum achieveable delay bound over previous constant-gain ones in the literature.

Solventless continuous flow homogeneous hydroformylation of 1-octene

The hydroformylation of 1-octene under continuous flow conditions is described. The system involves dissolving the catalyst, made in situ from [ Rh(acac)(CO)(2)] (acacH = 2,4- pentanedione) and [RMIM][TPPMS] ( RMIM = 1-propyl (Pr), 1-pentyl (Pn) or 1-octyl (O)-3-methyl imidazolium, TPPMS = Ph2P(3-C6H4SO3)), in a mixture of nonanal and 1-octene and passing the substrate, 1-octene, together with CO and H-2 through the system dissolved in supercritical CO2 (scCO(2)). [PrMIM][TPPMS] is poorly soluble in the medium so heavy rhodium leaching (as complexes not containing phosphine) occurs in the early part of the reaction. [PnMIM][ PPMS] affords good rates at relatively low catalyst loadings and relatively low overall pressure (125 bar) with rhodium losses <1 ppm, but the catalyst precipitates at higher catalyst loadings, leading to lower reaction rates. [OMIM][ TPPMS] is the most soluble ligand and promotes high reaction rates, although preliminary experiments suggested that rhodium leaching was high at 5-10 ppm. Optimisation aimed at balancing flows so that the level within the reactor remained constant involved a reactor set up based around a reactor fitted with a sight glass and sparging stirrer with the CO2 being fed by a cooled head HPLC pump, 1-octene by a standard HPLC pump and CO/H-2 through a mass flow controller. The pressure was controlled by a back pressure regulator. Using this set up, [OMIM][ TPPMS] as the ligand and a total pressure of 140 bar, it was possible to control the level within the reactor and obtain a turnover frequency of ca. 180 h(-1). Rhodium losses in the optimised system were 100 ppb. Transport studies showed that 1-octene is preferentially transported over the aldehydes at all pressures, although the difference in mol fraction in the mobile phase was less at lower pressures. Nonanal in the mobile phase suppresses the extraction of 1-octene to some extent, so it is better to operate at high conversion and low pressure to optimise the extraction of the products relative to the substrate. CO and H2 in the mobile phase also suppress the extraction effciency by as much as 80%.

Towards variance-matrix characterization of complementarity relations in a continuous variable system

We discuss complementarity relations in a bipartite continuous variable system. Building up from the work done on discrete d-dimensional systems, we prove that for symmetric two-mode states, quantum complementarity relations can be put in a simple relation with the elements of the variance matrix. When this condition is not satisfied, such a connection becomes non-trivial. Our investigation is the first step towards an operative characterization of the complementarity in a scenario that has not been investigated so far.

Continuous variation in the pattern of marine v. freshwater foraging in brown trout Salmo trutta L. from Loch Lomond, Scotland

Carbon stable-isotope analysis showed that individual brown trout Salmo trutta in Loch Lomond adopted strategies intermediate to that of freshwater residency or anadromy, suggesting either repeated movement between freshwater and marine environments, or estuarine residency. Carbon stable-isotope (delta C-13) values from Loch Lomond brown trout muscle tissue ranged from those indicative of assimilation of purely freshwater-derived carbon to those reflecting significant utilization of marine-derived carbon. A single isotope, two-source mixing model indicated that, on average, marine C made a 33% contribution to the muscle tissue C of Loch Lomond brown trout. Nitrogen stable isotope, delta N-15, but not delta C-13 was correlated with fork length suggesting that larger fish were feeding at a higher trophic level but that marine feeding was not indicated by larger body size. These results are discussed with reference to migration patterns in other species. (c) 2008 The Authors Journal compilation (c) 2008 The Fisheries Society of the British Isles.

Two-stage mixed discrete-continuous identification of radial basis function (RBF) neural models for nonlinear systems

The identification of nonlinear dynamic systems using radial basis function (RBF) neural models is studied in this paper. Given a model selection criterion, the main objective is to effectively and efficiently build a parsimonious compact neural model that generalizes well over unseen data. This is achieved by simultaneous model structure selection and optimization of the parameters over the continuous parameter space. It is a mixed-integer hard problem, and a unified analytic framework is proposed to enable an effective and efficient two-stage mixed discrete-continuous; identification procedure. This novel framework combines the advantages of an iterative discrete two-stage subset selection technique for model structure determination and the calculus-based continuous optimization of the model parameters. Computational complexity analysis and simulation studies confirm the efficacy of the proposed algorithm.