No description available.

No description available.

No description available.

In this thesis a detailed examination is carried out into the role and impact of the volume of trade in UK futures markets. While the success of a market may be judged by the number of investors that it attracts, how does the behaviour of individuals influence such key variables as price volatility and the cost of trading? The empirical work carried out here allows a unique appreciation of issues that have important implications for policy makers, investors and the practitioner. Motivated by a desire to understand whether volatility is destabilising or a reflection of fundamental factors, as well as the nature of the distribution of price returns, the relationship between volume and price movements is investigated in detail. The preliminary analysis suggests an important role for the flow of information which is confirmed by the rigorous testing of Anderson's (1996) specification of the Mixture of Distributions Hypothesis. The exploitation of this model allows an in-depth analysis of the information process including the identification of the informed and uninformed components of volume. There is also an investigation into the possibility that the volume statistic itself has an informative value. Using the Blume et al. (1994) approach the results suggest that, for a variety of futures contracts, the markets show a high degree of information dispersion. The need to attract investors has never been more acute than in today's competitive financial environment. It is therefore important to obtain a good appreciation of the relationship between volume and the cost of trading. This thesis includes a comprehensive intra-day study of the relation within a simultaneous econometric framework that exploits state-space models to investigate how markets react to unexpected levels of trading. The results question the dominance of inventory cost models and suggest that patterns of trade have become more predictable since contract inception, despite increases in volume.

Teleoperation systems are an important tool for performing tasks which require the sensori-motor coordination of an operator but where it is physically impossible for an operator to undertake such tasks in situ. The vast majority of these devices supply the operator with both visual and haptic sensory feedback in order that the operator can perform the task at hand as naturally and fluently as possible and as though physically present at the remote site. This thesis is concerned with overcoming the sensory limitations imposed by a fixed camera teleoperation system. The principal aim of this work is the extraction and redisplay of visual information to facilitate such a system. The thesis augments the Oxford teleoperation system with a virtual viewing module, where the operator is able to select his or her viewpoint and viewing direction onto the workcell by first tracking the locations of known objects in the workcell using a computer vision system, and then rendering them graphically on a display in front of the operator. This system, because the model-based object tracker is based around a Kalman filter, motivates the design of experiments to examine whether the operator's visuo-motor control loop maintains a state model of the manipulation process as in a Kalman filter. Experimental evidence is presented showing the latter to be false. A new operator model is then postulated using an adaptive gain controller, with the gain chosen to minimise the variance between desired and actual output. The experimental evidence supports this model. These findings support the hypothesis that the required bandwidth of the tracking filter is both i) sufficient that the tracker can robustly track hand manipulated objects and ii) matched to the visual needs of the operator.

Filtering has been shown successful in prediction from dynamically changing data. In this thesis, we perform case studies and comparison among three filters: Kalman filter, unscented Kalman filter and particle flow filter. We consider Kalman filter in the first chapter where we focus on studying the S&P model in a time-discrete dynamics with time-discrete observations for dividend yield and S&P returns. For this filtering problem, Kalman filter performs well only in the first few time steps. Since the S&P model we consider is nonlinear, we are motivated to apply nonlinear filters and use unscented Kalman filter. The key technique is to approximate non-Gaussian processes (non-linear models) by assigning the so-called sigma points (nonrandom) around the priori mean. We implement it on the S&P model in Chapter 2. We also implement unscented Kalman filter for a two-dimensional tumor growth model. Unscented Kalman filter works reasonably well for both models with capturing the trend and predicting the values. We consider the recently-developed particle flow filter in Chapter 3. Particle flow filter is a method of moving the particles by partial differential equations generated from proper chosen likelihood functions via the Bayes rule. By solving partial differential equations, one can construct an explicit dynamic model on how to move particles.In this chapter, we implement two models as in Chapter 2. One is the S&P model and the other is perturbed tumor growth model. We compare performance of particle flow filter and unscented Kalman filter for these two models.

Local active noise control systems aim to create zones of quiet at specific locations within a sound field. The created zones of quiet generally tend to be small, especially for higher frequencies, and are usually centred at the error sensors. For an observer to experience significant reductions in the noise, the error sensors therefore have to be placed relatively close to an observer’s ears, which is not always feasible or convenient. Virtual sensing methods have been proposed to overcome these problems that have limited the scope of successful local active noise control applications. These methods require non-intrusive sensors that are placed remotely from the desired locations of maximum attenuation. These non-intrusive sensors are used to provide an estimate of the sound pressures at these locations, which can then be minimised by a local active noise control system. This effectively moves the zones of quiet away from the physical locations of the transducers to the desired locations of maximum attenuation, such as a person’s ears. A number of virtual sensing algorithms have been proposed previously. The difference between these algorithms is the structure that is assumed to compute an estimate of the virtual error signals. The question now arises as to whether there is an optimal structure that can be used to solve the virtual sensing problem, which amounts to a linear estimation problem. It is well-known that the Kalman filter provides an optimal structure for solving such problems. An optimal solution to the virtual sensing problem is therefore derived in this thesis using Kalman filtering theory. The proposed algorithm is implemented on an acoustic duct arrangement to demonstrate its effectiveness. The presented experimental results indicate that the zone of quiet was effectively moved away from the physical sensor towards the desired location of maximum attenuation. The previously proposed virtual sensing algorithms have been developed with the aim to create zones of quiet at virtual locations that are assumed spatially fixed within the sound field. Because an observer is very likely to move their head, the desiredlocations of the zones of quiet are generally moving through the sound field rather than being spatially fixed. For effective control, a local active noise control system incorporating a virtual sensing method thus has to be able to create moving zones of quiet that track the observer’s ears. A moving virtual sensing method is therefore developed in this thesis that can be used to estimate the error signals at virtual locations that are moving through the sound field. It is shown that an optimal solution to the moving virtual sensing problem can be derived using Kalman filtering theory. A practical implementation of the developed algorithm is combined with an adaptive feedforward control algorithm and implemented on an acoustic duct arrangement. The presented experimental results illustrate that a narrowband moving zone of quiet that tracks the desired location of maximum attenuation has successfully been created. / http://proxy.library.adelaide.edu.au/login?url= http://library.adelaide.edu.au/cgi-bin/Pwebrecon.cgi?BBID=1291123 / Thesis (Ph.D.) -- University of Adelaide, School of Mechanical Engineering, 2007.

The multi-scale mixed finite element method (MsMFEM) discussed in this work uses a two-scale approach, where the solutions to independent local flow problems on the fine grid capture the fine-scale variations of the reservoir model, while the coarse grid equations appropriately assimilate this information in the global solution. Temporal changes in porous media flow are relatively moderate when compared to the spatial variations in the reservoir. Hence, approximate global solutions by adaptively solving these local flow problems can be obtained with significant savings in computational time. The ensemble Kalman filter, used for real-time updating of reservoir models, can thus be coupled with the MsMFEM-streamline simulator to speed up the historymatching process considerably.

Abstract A feasible technique, using carrier-phase data from GPS and CCD camera, is presented to identify ground target location as well as azimuth angle of a low altitude aircraft/helicopter without using any gyroscope measurements; the baseline vector can also be identified using GPS compass. The ground target¡¦s image is extracted from background and recorded by image processing technique. By integrating ground target¡¦s location and the recorded GPS data, the designated states can be estimated by using extended Kalman filter technique. Basically, the extended Kalman filter does the state estimation job, and it¡¦s a nonlinear measurement process. By processing these time update and measurement update, the integer ambiguity as well as azimuth angle can be determined. The proposed GPS compass system consists of three componets : pointer, sensor, and controller. By using carrier-phase data from two GPS receivers, we can compute the baseline vector, whose length is equal to one meter, and achieve the direction accuracy within one degree. The integer ambiguity number is resolved by rotating the baseline vector; the conventional antenna swapping technique is a special case of the proposed method. Therefore, the GPS compass may replace these magnetic compass or gyroscope used in navigation system. By continuously snapping ground target image using CCD camera and utilizing the GPS receivers, the coordinate of the ground target can be identified. Simulation justifies the feasibility of the proposed scenario. Simulation has shown that the estimation errors for stationary and traveling with constant velocity ground targets are within 1.8 m and 6 m, respectively.

Kalman filter addresses an estimation problem defined by two models: the signal model and the observation model. In this thesis, the signal model is obtained from a ratio of the defected and clean pictures in frequency domain. The observation model is built for an additive measurement noise from electronic sampling. The statistics of the generating noise for the signal model is important in Kalman filtering. The focus of this thesis is to derive the variance of the generating noise in the middle band for the signal model. By this derived variance, the Kalman filter is thus possible to be applied to estimate the optical transfer function for a defected imaging system in the future.