Search

This thesis explores modernist narrative art embodied in modernist style of constructing narrative space. Within Chatman’s conceptual framework, narrative space can be divided into story-space (settings and characters) and discourse-space (focus of spatial attention). In a corpus-stylistic approach, the structuration of the story-space in The Mill on the Floss, The Good Soldier and To the Lighthouse is examined. The findings show that modernist tendency to deemphasise particularity of place shapes a narrative design of spatial detachment. In consequence, the establishment of settings in early modernist fiction is generally sketchy, but sometimes spatially informative. This is a mixed character. By contrast, settings in classic modernist fiction are symbolic of viewers’ psychological states, a clear manifestation of a modernist interest in characters’ interiority. To further trace the style change from early modernism to high modernism, a cross-disciplinary model for character analysis and a cross-axial model for the examination of discourse-space have been constructed. They help detect some similarities and dissimilarities between early and classic modernist styles of spatialisation. As a whole, this thesis has two features. First, it applies corpus stylistic methods to inform cognitive narratological interpretation. Second, it resorts to visualisation as an attempt at a multi-modal study of narrative space.

In this thesis a new adaptive control method, called Iterative Tuning in the Frequency Domain (FD-IT), is proposed for Active Noise and Vibration Control (ANVC). This approach is a gradient based self-tuning method which completely relies on analysis of the frequency response of system dynamics and the spectrum of signals. The new method is based on a new gradient estimation theory in the frequency domain. In this theory the gradient of the output spectrum with respect to controller parameters is expressed with the frequency response of dynamics and the spectrum of signals. When the performance gradient with respect to controller parameters can be expressed as some function of the signals’ spectrum, it can be computed out completely in the frequency domain. Similar to audio compression, when the system’s signals contain few frequencies, the computation of performance gradient can be greatly simplified by making ”partial modelling” with respect to those frequencies. According to the proposed theory, the new iterative tuning method, i.e., FD-IT, is developed for ANVC problems with periodic disturbances. It can tune the feedback/feed-forward controllers simultaneously with one experiment per iteration except some extra experiments for initial tuning. It covers both Single Input Single Output (SISO) and Multiple Input Multiple Output (MIMO) systems. Furthermore, it can be extended to nonlinear systems as well. Some issues about the implementation of the iterative method are discussed. Through the comparison with some other popular active control methods in ANVC, the advantages of the new method, including: the flexibility in selecting controllers, the simplicity in control structure, and the convenience in implementation, are emphasized. The effectiveness and robustness of the proposed iterative tuning method are tested through simulated SISO and MIMO Linear Time Invariant (LTI) systems. Two simulated nonlinearities are used to illustrate the usefulness of the methods in nonlinear system as well. To show the practicability, the linear and nonlinear FD-ITs are implemented in an air-duct system with a PC-DSP based agent-architecture. All the results illustrate that FD-IT is an easy and effective approach to solve ANVC problems with periodic disturbances.

Abundant experimental evidence shows that the changes in Reynolds stresses caused by streamline curvature are an order of magnitude larger than would be expected from the explicit curvature-dependent terms appearing in the Reynolds-stress transport equations. Turbulence models developed for simple undistorted shear layers frequently fail to predict this observed changes. In the present work, Bradshaw's one-equation model of turbulence with an algebraic length scale has been used as the base model for turbulence modelling. Two improved schemes have been adopted for turbulent length scale. These are: (1) the algebraic length scale is modified by using the curvature-correction scheme proposed by Bradshaw (1973); (2) a transport equation for the turbulent length scale itself, which is modified by adding an extra curvature-dependent term, is developed to replace the algebraic one.

This thesis consists of two parts: 1) a result of the rationality of channel capacity for the M -ary (d, k) constraint, and 2) a software package which has been designed to aid in the development of efficient, high density channel encoders for M -ary (d, k) constrained recording code. It has been written for the platforms of IBM PC or PC compatible and Sun Workstations. The designed software in this thesis involves three steps. The first step is to calculate the capacity of a RLL code after the user has chosen (M,d,k) . The channel capacity specifies the maximum ratio of information ( p ) bits/codeword ( q ) bits achievable in implementing these codes and gives options of code choice for any specific task. The second step is to find a code efficiency (eta) = R/C by choosing p and q , and display the state transition matrix of finite state transition diagram and finite state transition diagram of the code. The last step is to aid the user in the generation of a code, by using state splitting algorithm of R. Adler, D. Coppersmith and M. Hassner [10].