Sie sind auf Seite 1von 5

Research Seminar

Date and Time: Venue: Wednesday, 5 May 2010, 13:00-15:30 Lower Peter Jost Lecture Theatre

Programme
12:45 13:00 13:00 13:30 Tea and Coffee

Dr Gabor Barton (Research Institute for Sport and Exercise Sciences) Advanced applications of biomechanics: interpretation of data supported by neural networks, and neuromuscular rehabilitation aided by virtual reality

13:30 13:50

Dr Gareth Bradley Synchrotron diffraction measurement of residual stresses

13:50-14:10

Dr Christian Matthews Engineering applications of System Identification and Robust control

14:10 14:30

Coffee Break

14:30-14:45 14:45 15:00

Mr Wahyu Satiawan Pulse width modulation of a dual-inverter supplied multi-phase motor drive Mr Barra Touray Energy-efficient algorithms for WSNs in environmental monitoring

15:00 15:15 15:15 15:30

Mr Justin Su Inverse analysis of material parameters based on indentation tests Mr Wayan Widhiada Design of a novel robotic gripper for the dexterous assembly of compliant elements

15:30

Dr James Ren, Discussion and Closing Remarks

Advanced applications of biomechanics: interpretation of data supported by neural networks, and neuromuscular rehabilitation aided by virtual reality Gabor Barton Major efforts are still being made to cover the fundamental methodological issues of biomechanical measurements but in the first decade of the 21st century the emerging need to answer the real questions has been met with advanced tools supporting data processing and interpretation of results. Two selected areas where some considerable progress has been made are the integration of large quantities of biomechanical data leading to practically meaningful interpretations which can support clinical decision making, and the advanced revisiting of biofeedback in the form of complex real-time reactive environments used to improve movement control. IBMs Deep Blue chess computer beating Gary Kasparov in 1997 was described by him as quantity becoming quality referring to the computers ability to analyse 200 million chess positions a second. Considering that megabytes of multi channel data can be easily collected in a typical biomechanical experiment, the challenge is to turn the massive quantities of data to information which is compatible with the qualitative evaluation of human experts. Artificial neural networks are mathematical simulations of the biological nervous system and are well suited to aid processing of biomechanical data (Chau, 2001). Among other uses, artificial neural networks can be used to classify patients with movement problems (Barton et al., 1997), to visualise multi-dimensional data in simplified formats (Barton et al., 2007), and to calculate the deviation from normality (Barton et al., 2003; 2010). Avatar the movie has brought the idea of virtual reality to the masses in the entertainment industry but virtual reality has strong research potential too. By placing an individual in a computer controlled reactive virtual environment, the pre-requisites of motor learning can be met efficiently. Repetition of the motor task dictated by the environment leads to gradual improvement in response to the feedback provided instantaneously in a motivating and often entertaining setting. Integrated super systems (e.g. CAREN system, MotekMedical, The Netherlands) enable accurate biomechanical measurement of movement together with means of influencing vision, proprioception and the vestibular system. Virtual rehabilitation can lead to improved movement function and the effects are often stronger than on the real task (Holden, 2005). Neuroplasticity of the intact or damaged central nervous system evidenced by functional MRI (You et al., 2005) is the underlying mechanism of improving movement control of the core (pelvis and trunk) leading to proximal stability (Barton et al., 2009) which is the pre-requisite of distal mobility.

Pulse width modulation of a dual-inverter supplied multi-phase motor drive Abstract I N. Wahyu Satiawan Two approaches can be considered to solve the problem of the limitation of the rating of the power electronics devices in medium/high power drive applications, either by increasing the number of phases of the motor, or by increasing the number of levels of the inverter. In the recent time, attempts to combine these two methods have led to the development of multi-level multi-phase VSIs. Dual-inverter supply structure is a relatively new breed of the multi-level inverter. In this topology, the neutral point of the machine is opened, and the stator windings are supplied from both sides. This inverter topology is expected to give a better output although more complicated PWM algorithms are also required. In this research, a five-phase open-end winding machine is fed by two-level inverter in each side. An equal dc voltage source is applied for both inverters, yields 1024 switching state combinations which are mapped in 211 locations. Two SVPWM schemes have been successfully developed. Both schemes utilise two five-phase two-level space-vector modulators, however the way to feed the inverters are different. An equally shared reference voltage for both inverters is applied in the first scheme. While in the second scheme, the distribution of the reference is governed by certain constraints based on the amplitude of the reference voltage. The simulation results show that the first scheme produces outputs identical to conventional fivephase two-level inverter. While the second scheme enables multi-level operation which is represented by more varied number of level in the phase output voltages. More vectors are used by modulator in the second scheme instead of only 22 vectors are used in the first scheme. As a result, better quality of the output voltages is achieved, indicated by the lower value of the total harmonic distortion (THD). As consequence, a significant improvement of the output voltage is obtained.

Energy-efficient algorithms for WSNs in environmental monitoring Barra Touray Wireless Sensor Network (WSN) is a network of spatially distributed autonomous devices that depend on their sensors to monitor environmental or physical conditions. There are various network models. The Random Network model assume that the connection between nodes occur randomly which is not the case in most real world networks. The small world network model is a model where nodes in a cluster are linked to other clusters by few nodes in between. However, this model fails to consider preferential attachment resulting to a new model called Scale-Free by Barabasi and Albert. So in, order to effectively model an energy-efficient scheduling algorithm, one need to also model the wireless sensor network topologies and not only rely on the assumption that WSN are random networks. It is proven that most real world networks exhibit other features that differentiate them from being purely random networks. The success for modelling energy-efficient scheduling algorithm will depend on knowing those features of the topology of the WSN. Knowing these, will enable one to use it to its advantage or avoid it from being a bottleneck in the modelling. This paper will discuss the outcome of the research work in developing a statistical model for the WSN based on the Small World Network topology suitable for applications within environmental monitoring. The model will be simulated using the OPNET modelling software package and their performance will be evaluated against standard network parameters.

Design of a novel robotic gripper for the dexterous assembly of compliant elements
Wayan Widhiada

A robot gripper is an essential element of the robotic system and it is tailored to suit the application in hand. A well designed gripper can simplify the overall assembly, improving overall reliability system, and reduce implementation costs. Grippers have been widely used for automated manufacturing, assembly and packaging, etc. Unfortunately, very little attention has been given to the research and design of grippers for high speed assembly of relatively small compliant elements A three fingered robot gripper has been and modelled in Solidworks. A novel feature of this gripper is the inclusion of a prismatic sliding element at the end of each finger to facilitate the grippers handling of both large and small compliant components typically found in the assembly of a domestic gas regulator. An 11 DOFs kinematical model of the three fingers robot gripper is presented. The first body is the palm connected with the proximal phalanges of the fingers. Each finger is linked to the palm at different points and is modelled as an independent kinematical chain.

PID control is applied to control the gross motions positions, angular velocities and angular accelerations in the joint angle of the fingers. By choosing a suitable value of the three constant gains in the PID controller algorithm, the controller can provide control action designed for specific process requirements. The response of the controller can be described in terms of the responsiveness of the controller to an error, the degree to which the controller overshoots the set point and the degree of system oscillation.
Keywords: Three Fingers Gripper, Prismatic Sliding Element, PID Control

Das könnte Ihnen auch gefallen