Neurofuzzy design and model construction of nonlinear dynamical processes from data

A common problem in many data based modelling algorithms such as associative memory networks is the problem of the curse of dimensionality. In this paper, a new two-stage neurofuzzy system design and construction algorithm (NeuDeC) for nonlinear dynamical processes is introduced to effectively tackle this problem. A new simple preprocessing method is initially derived and applied to reduce the rule base, followed by a fine model detection process based on the reduced rule set by using forward orthogonal least squares model structure detection. In both stages, new A-optimality experimental design-based criteria we used. In the preprocessing stage, a lower bound of the A-optimality design criterion is derived and applied as a subset selection metric, but in the later stage, the A-optimality design criterion is incorporated into a new composite cost function that minimises model prediction error as well as penalises the model parameter variance. The utilisation of NeuDeC leads to unbiased model parameters with low parameter variance and the additional benefit of a parsimonious model structure. Numerical examples are included to demonstrate the effectiveness of this new modelling approach for high dimensional inputs.

Data based constructive identification - overcoming the curse of dimensionality

The modelling of a nonlinear stochastic dynamical processes from data involves solving the problems of data gathering, preprocessing, model architecture selection, learning or adaptation, parametric evaluation and model validation. For a given model architecture such as associative memory networks, a common problem in non-linear modelling is the problem of "the curse of dimensionality". A series of complementary data based constructive identification schemes, mainly based on but not limited to an operating point dependent fuzzy models, are introduced in this paper with the aim to overcome the curse of dimensionality. These include (i) a mixture of experts algorithm based on a forward constrained regression algorithm; (ii) an inherent parsimonious delaunay input space partition based piecewise local lineal modelling concept; (iii) a neurofuzzy model constructive approach based on forward orthogonal least squares and optimal experimental design and finally (iv) the neurofuzzy model construction algorithm based on basis functions that are Bézier Bernstein polynomial functions and the additive decomposition. Illustrative examples demonstrate their applicability, showing that the final major hurdle in data based modelling has almost been removed.

The application of Prony’s method to time-domain waveform data modeling in the presence of noise

In this paper,the Prony's method is applied to the time-domain waveform data modelling in the presence of noise.The following three problems encountered in this work are studied:(1)determination of the order of waveform;(2)de-termination of numbers of multiple roots;(3)determination of the residues.The methods of solving these problems are given and simulated on the computer.Finally,an output pulse of model PG-10N signal generator and the distorted waveform obtained by transmitting the pulse above mentioned through a piece of coaxial cable are modelled,and satisfactory results are obtained.So the effectiveness of Prony's method in waveform data modelling in the presence of noise is confirmed.

Two methods for modelling the propagation of terahertz radiation in a layered structure

Modelling the interaction of terahertz(THz) radiation with biological tissueposes many interesting problems. THzradiation is neither obviously described byan electric field distribution or anensemble of photons and biological tissueis an inhomogeneous medium with anelectronic permittivity that is bothspatially and frequency dependent making ita complex system to model.A three-layer system of parallel-sidedslabs has been used as the system throughwhich the passage of THz radiation has beensimulated. Two modelling approaches havebeen developed a thin film matrix model anda Monte Carlo model. The source data foreach of these methods, taken at the sametime as the data recorded to experimentallyverify them, was a THz spectrum that hadpassed though air only.Experimental verification of these twomodels was carried out using athree-layered in vitro phantom. Simulatedtransmission spectrum data was compared toexperimental transmission spectrum datafirst to determine and then to compare theaccuracy of the two methods. Goodagreement was found, with typical resultshaving a correlation coefficient of 0.90for the thin film matrix model and 0.78 forthe Monte Carlo model over the full THzspectrum. Further work is underway toimprove the models above 1 THz.

Using modelling tools to optimise the calculation and subsequent reduction of cable distortion parameters with application to oil well logging systems

This paper discusses how the use of computer-based modelling tools has aided the design of a telemetry unit for use with oil well logging. With the aid of modern computer-based simulation techniques, the new design is capable of operating at data rates of 2.5 times faster than previous designs.

Physics-based modelling for remote sensing of snow

Remote sensing is the only practicable means to observe snow at large scales. Measurements from passive microwave instruments have been used to derive snow climatology since the late 1970’s, but the algorithms used were limited by the computational power of the era. Simplifications such as the assumption of constant snow properties enabled snow mass to be retrieved from the microwave measurements, but large errors arise from those assumptions, which are still used today. A better approach is to perform retrievals within a data assimilation framework, where a physically-based model of the snow properties can be used to produce the best estimate of the snow cover, in conjunction with multi-sensor observations such as the grain size, surface temperature, and microwave radiation. We have developed an existing snow model, SNOBAL, to incorporate mass and energy transfer of the soil, and to simulate the growth of the snow grains. An evaluation of this model is presented and techniques for the development of new retrieval systems are discussed.

Grey-box radial basis function modelling

A fundamental principle in data modelling is to incorporate available a priori information regarding the underlying data generating mechanism into the modelling process. We adopt this principle and consider grey-box radial basis function (RBF) modelling capable of incorporating prior knowledge. Specifically, we show how to explicitly incorporate the two types of prior knowledge: (i) the underlying data generating mechanism exhibits known symmetric property, and (ii) the underlying process obeys a set of given boundary value constraints. The class of efficient orthogonal least squares regression algorithms can readily be applied without any modification to construct parsimonious grey-box RBF models with enhanced generalisation capability.

Modelling and control of Hammerstein system using B-spline approximation and the inverse of De Boor algorithm

In this article a simple and effective controller design is introduced for the Hammerstein systems that are identified based on observational input/output data. The nonlinear static function in the Hammerstein system is modelled using a B-spline neural network. The controller is composed by computing the inverse of the B-spline approximated nonlinear static function, and a linear pole assignment controller. The contribution of this article is the inverse of De Boor algorithm that computes the inverse efficiently. Mathematical analysis is provided to prove the convergence of the proposed algorithm. Numerical examples are utilised to demonstrate the efficacy of the proposed approach.

Issues in the assimilation of the sources of a trace gas by inverse modelling

The problem of reconstructing the (otherwise unknown) source and sink field of a tracer in a fluid is studied by developing and testing a simple tracer transport model of a single-level global atmosphere and a dynamic data assimilation system. The source/sink field (taken to be constant over a 10-day assimilation window) and initial tracer field are analysed together by assimilating imperfect tracer observations over the window. Experiments show that useful information about the source/sink field may be determined from relatively few observations when the initial tracer field is known very accurately a-priori, even when a-priori source/sink information is biased (the source/sink a-priori is set to zero). In this case each observation provides information about the source/sink field at positions upstream and the assimilation of many observations together can reasonably determine the location and strength of a test source.

Evaluating a new LISFLOOD-FP formulation with data from the summer 2007 floods in Tewkesbury, UK

Two-dimensional flood inundation modelling is a widely used tool to aid flood risk management. In urban areas, the model spatial resolution required to represent flows through a typical street network often results in an impractical computational cost at the city scale. This paper presents the calibration and evaluation of a recently developed formulation of the LISFLOOD-FP model, which is more computationally efficient at these resolutions. Aerial photography was available for model evaluation on 3 days from the 24 to the 31 of July. The new formulation was benchmarked against the original version of the model at 20 and 40 m resolutions, demonstrating equally accurate simulation, given the evaluation data but at a 67 times faster computation time. The July event was then simulated at the 2 m resolution of the available airborne LiDAR DEM. This resulted in more accurate simulation of the floodplain drying dynamics compared with the coarse resolution models, although maximum inundation levels were simulated equally well at all resolutions tested.

Integration of remote sensing and GIS for modelling flash floods in Wadi Hudain catchment, Egypt

This paper presents a new approach to modelling flash floods in dryland catchments by integrating remote sensing and digital elevation model (DEM) data in a geographical information system (GIS). The spectral reflectance of channels affected by recent flash floods exhibit a marked increase, due to the deposition of fine sediments in these channels as the flood recedes. This allows the parts of a catchment that have been affected by a recent flood event to be discriminated from unaffected parts, using a time series of Landsat images. Using images of the Wadi Hudain catchment in southern Egypt, the hillslope areas contributing flow were inferred for different flood events. The SRTM3 DEM was used to derive flow direction, flow length, active channel cross-sectional areas and slope. The Manning Equation was used to estimate the channel flow velocities, and hence the time-area zones of the catchment. A channel reach that was active during a 1985 runoff event, that does not receive any tributary flow, was used to estimate a transmission loss rate of 7·5 mm h−1, given the maximum peak discharge estimate. Runoff patterns resulting from different flood events are quite variable; however the southern part of the catchment appears to have experienced more floods during the period of study (1984–2000), perhaps because the bedrock hillslopes in this area are more effective at runoff production than other parts of the catchment which are underlain by unconsolidated Quaternary sands and gravels. Due to high transmission loss, runoff generated within the upper reaches is rarely delivered to the alluvial fan and Shalateen city situated at the catchment outlet. The synthetic GIS-based time area zones, on their own, cannot be relied on to model the hydrographs reliably; physical parameters, such as rainfall intensity, distribution, and transmission loss, must also be considered.

An ocean modelling and assimilation guide to using GOCE geoid products

We review the procedures and challenges that must be considered when using geoid data derived from the Gravity and steady-state Ocean Circulation Explorer (GOCE) mission in order to constrain the circulation and water mass representation in an ocean 5 general circulation model. It covers the combination of the geoid information with timemean sea level information derived from satellite altimeter data, to construct a mean dynamic topography (MDT), and considers how this complements the time-varying sea level anomaly, also available from the satellite altimeter. We particularly consider the compatibility of these different fields in their spatial scale content, their temporal rep10 resentation, and in their error covariances. These considerations are very important when the resulting data are to be used to estimate ocean circulation and its corresponding errors. We describe the further steps needed for assimilating the resulting dynamic topography information into an ocean circulation model using three different operational fore15 casting and data assimilation systems. We look at methods used for assimilating altimeter anomaly data in the absence of a suitable geoid, and then discuss different approaches which have been tried for assimilating the additional geoid information. We review the problems that have been encountered and the lessons learned in order the help future users. Finally we present some results from the use of GRACE geoid in20 formation in the operational oceanography community and discuss the future potential gains that may be obtained from a new GOCE geoid.

View-based modelling of human visual navigation errors

View-based and Cartesian representations provide rival accounts of visual navigation in humans, and here we explore possible models for the view-based case. A visual “homing” experiment was undertaken by human participants in immersive virtual reality. The distributions of end-point errors on the ground plane differed significantly in shape and extent depending on visual landmark configuration and relative goal location. A model based on simple visual cues captures important characteristics of these distributions. Augmenting visual features to include 3D elements such as stereo and motion parallax result in a set of models that describe the data accurately, demonstrating the effectiveness of a view-based approach.

Measuring the discrepancy between fecundity and lifetime reproductive success in a pollinating fig wasp

Lifetime reproductive success in female insects is often egg- or time-limited. For instance in pro-ovigenic species, when oviposition sites are abundant, females may quickly become devoid of eggs. Conversely, in the absence of suitable oviposition sites, females may die before laying all of their eggs. In pollinating fig wasps (Hymenoptera: Agaonidae), each species has an obligate mutualism with its host fig tree species [Ficus spp. (Moraceae)]. These pro-ovigenic wasps oviposit in individual ovaries within the inflorescences of monoecious Ficus (syconia, or ‘figs’), which contain many flowers. Each female flower can thus become a seed or be converted into a wasp gall. The mystery is that the wasps never oviposit in all fig ovaries, even when a fig contains enough wasp females with enough eggs to do so. The failure of all wasps to translate all of their eggs into offspring clearly contributes to mutualism persistence, but the underlying causal mechanisms are unclear. We found in an undescribed Brazilian Pegoscapus wasp population that the lifetime reproductive success of lone foundresses was relatively unaffected by constraints on oviposition. The number of offspring produced by lone foundresses experimentally introduced into receptive figs was generally lower than the numbers of eggs carried, despite the fact that the wasps were able to lay all or most of their eggs. Because we excluded any effects of intraspecific competitors and parasitic non-pollinating wasps, our data suggest that some pollinators produce few offspring because some of their eggs or larvae are unviable or are victims of plant defences.

Generic unified modelling process for developing semantically rich, dynamic and temporal models

Models play a vital role in supporting a range of activities in numerous domains. We rely on models to support the design, visualisation, analysis and representation of parts of the world around us, and as such significant research effort has been invested into numerous areas of modelling; including support for model semantics, dynamic states and behaviour, temporal data storage and visualisation. Whilst these efforts have increased our capabilities and allowed us to create increasingly powerful software-based models, the process of developing models, supporting tools and /or data structures remains difficult, expensive and error-prone. In this paper we define from literature the key factors in assessing a model’s quality and usefulness: semantic richness, support for dynamic states and object behaviour, temporal data storage and visualisation. We also identify a number of shortcomings in both existing modelling standards and model development processes and propose a unified generic process to guide users through the development of semantically rich, dynamic and temporal models.