J.P. Norton

Predictive self-tuning control by parameter bounding and worst-case design

Abstract The computation of bounds on the parameters of a plant model allows worst-case control synthesis, taking account of the uncertainty in the model. This paper introduces such a control scheme: predictive bounding control. The scheme contrasts with existing self-tuning control methods which base control synthesis on a nominal plant model. Parameter bounding also permits detection of abrupt plant changes and adaptive tracking of time-varying plant characteristics by suitable choice of bounds on plant-model output error and plant-parameter increments. Estimation and control are closely integrated and the control computation can compromise between reducing the model uncertainty and reducing predicted output error. Simulation examples show the excellent performance of predictive bounding control.

Structure selection for bounded-parameter models: consistency conditions and selection criterion

Strong-consistency conditions for structure selection in bounded-parameter models are studied. A certain robust selection criterion, based on the volume of the exact parameter-bounding polytope, is proposed for linear regression models. The effectiveness of the polytope volume criterion is demonstrated on a model nonlinear in its variables but linear in its parameters. Its strong consistency is proved for a large class of noise distributions. The usual assumptions on the noise, namely independence, constant variance, or martingale difference properties, need not be made, but asymptotic independence is assumed. >

Sensitivity analysis for assessing the behaviour of a landscape-based sediment source and transport model

Abstract Widespread degradation of aquatic habitat and water quality has occurred since European settlement in Australia. Repairing this degradation is expensive and hence on-ground management needs to be carefully focussed. The Sediment River Network model, SedNet, used to estimate (spatially) the sources and transport of sediment at catchment scales provides a potentially useful tool to assist land managers in focusing this work. The complete model, while broadly applied has not been systematically tested to assess its sensitivity to its various model components. This paper describes sensitivity assessment to improve understanding of the model, with the aim of prioritising data acquisition and improving the structure and parameterisation of the model where necessary. It was found that the SedNet model was most sensitive to perturbations in its hydrologic parameters across a variety of scales and sites. Because of this it is important that ongoing model development activities focus on how to effectively represent hydrologic processes in the model.

Application of system identification techniques to aircraft gas turbine engines

Abstract A variety of system identification techniques are applied to the modelling of aircraft gas turbine dynamics. The motivation behind the study is to improve the efficiency and cost-effectiveness of system identification techniques currently used in the industry. Three system identification approaches are outlined in this paper. They are based upon: multisine testing and frequency-domain identification, time-varying models estimated using extended least squares with optimal smoothing, and multiobjective genetic programming to select model structure.

System identification strategies applied to aircraft gas turbine engines

Abstract A variety of system identification techniques are applied to the derivation of models of aircraft gas turbine dynamics. The motivation behind the study is to improve the efficiency and cost-effectiveness of system identification techniques currently used in the industry. Four system identification approaches are outlined in this paper. They are based upon: identification using ambient noise only data, multisine testing and frequency-domain identification, time-varying models estimated using extended least squares with optimal smoothing, and multiobjective genetic programming to select model structure.

Parameter-Bounding Algorithms for Linear Errors-in-Variables Models

Abstract The static and dynamic cases of parameter bounding for errors-in-variables models are discussed and their differences clarified. Algorithms to calculate parameter bounds for such models are presented. A search method and polytope and ellipsoid bounding are considered. Methods are given for finding orthants of the parameter space which contain no feasible parameter vectors and can be discarded, making the bounding algorithm faster. Simulation examples illustrate the problems and the algorithms.

A new inverse method for trace gas flux estimation: 2. Application to tropospheric CFCl3 fluxes

An initial application of a new inverse method for the estimation of flux strengths of long-lived atmospheric trace gases is presented. CFCl3 is studied using the Australian National Universitys Chemical Transport Model. Unit-pulse responses are derived from the model, and used to identify a time-varying state-space model of tropospheric CFCl3. This in turn is used in a Kaiman filter to perform two input-estimation studies. The first uses model generated measurements to estimate known flux strengths. This demonstrates the robustness of the method, although it is found that instantaneous stratospheric loss rates are not well-estimated using only surface concentration measurements. Emissions however, are robustly estimated. A time-aver aged tropospheric lifetime can be estimated however, with an accuracy of ±5 years. Thirteen years of Atmospheric Lifetime Experiment Global Atmospheric Gases Experiment CFCl3 measurements are used in the second flux estimation experiment. The estimated fluxes though are outside known physical limits for CFCl3, and it is concluded that either a more accurate/appropriate transport model, or more measurement locations, are needed to obtain useful information regarding regional tropospheric CFCl3 fluxes.

A new inverse method for trace gas flux estimation : 1. State-space model identification and constraints

A new method for estimating time-varying fluxes of atmospheric trace gases using an atmospheric transport model and observed concentrations is presented. Specifically Kaiman filtering is used to estimate inputs from a state-space model identified using unit-pulse response functions from a transport model. The method is new in that no assumptions about initial concentrations in the model are required, although this in turn means that all flux processes must be explicitly modeled as inputs linearly related to concentrations. This also means that at least one extra measuring-site or other measurement variable (e.g. a linear combination of emissions) than the number of input-fluxes being estimated, is required to ensure a stable Kaiman filter.

Structural zeros in the modal matrix and its inverse

It is shown that the positions of the zeros in the modal matrix of a system, and its inverse, which result from zeros in the state matrix, can be predicted by computing the corresponding reachability matrix.

Outliers in bound-based state estimation and identification

In the observation-based computation of bounds on the parameters or state of a model with specified bounds on output error or process and observation respectively, outliers may be defined as observations giving rise to clashes of the parameter or state bounds. Possible ways to detect and locate outliers in this context are considered. Static and dynamic problems are distinguished. Upper bounds, independent of the record length, on the numbers of isolated outliers or k-member outlier sets are given. Two possible approaches for locating outliers in the dynamic case are suggested. An efficient algorithm is presented for locating parameter or state bounds responsible for clashes.<<ETX>>