Overview of ISVR CAV Workshop 2011

Transcription

1 Overview of ISVR CAV Workshop 2011

2 New Faculty Structure FACULTY ST UDENT S (FTE) ACADEMIC STAFF (FT E) NON- ACADEMIC ST A FF (FT E) TURNOVER ( M) Hum anities Business & Law Includes: Management and Law Engineering & the Environment Includes: ISVR, SES & CEE Natural Sciences Includes: SOES, Chemistry and Biological Sciences Health Sciences Medicine Physical & Applied Science Includes: ORC, ECS and Phy sics & Astronomy Social & Hum an Sciences Includes: Social Sciences, Education, Geography, Mathematics, SSSRI and Psychology

3 Calculating the ISVR balance sheet

4 ISVR Structure Director: Jeremy Astley Deputy HoS Research: Tim Leighton Deputy HoS Education: Neil Ferguson Research Groups Dynamics: Brian Mace Fluid Dynamics & Acoustics Tim Leighton Human Sciences Mike Griffin Signal Processing & Control Group Bob Allen Clinics and Consulting Units South of England Cochlear Implant Centre Julie Brinton/Julie Eyles ISVR Consulting Malcolm Smith

5 South of England Cochlear Implant Centre South of England Cochlear Implant Centre Turnover 5,000,000 4,500,000 4,000,000 3,500,000 3,000,000 2,500,000 2,000,000 1,500,000 1,000, , / / / / / / / /09

6 Current Research Aeroacoustics Structural dynamics Signal processing Hearing aids Response to motion Bioacoustics Ultrasonics Active control Cochlear implants Health effects

7 External Context of Research Environmental Awareness Increased need for Hearing Prosthetics Noise in community Fuel efficient/lightweight vehicles Noise and vibration issues Technological Advances Ageing population Greater expectations Moore s law Implantability

8 Strategic Research Areas Vehicle noise and vibration Active systems Hearing aids and cochlear implants Ultrasonics Underwater Acoustics Signal processing Biodynamics and balance

9 Active Control of Vibrations using the Receptance Method Maryam Ghandchi Tehrani

10 Active Vibration Control for Marine Application Vibration Control for Marine Structures Prof. Steve Daley SPCG ISVR Centre for Research in Active Control

11 Active Vibration Control for Marine Application Broadband / Harmonic Vibration Low frequency Seaway Motion Conventional DG Set Raft Passive Isolation

12 Active Vibration Control for Marine Application Main marine application research started in late 1980 s Variety of sponsors, collaborators. and employers Maglev Raft Prime contractor GEC -Marconi Funded through DARPA /ONR ~ 10 year - $10m programme Prototype raft Control theory Low noise switching amplifier Novel electromagnet design DSP based controller hardware (186 processors)

13 Active Vibration Control for Marine Application Smart Spring Concept based on enhancing best passive design Original concept maintains dynamic features of Maglev Raft design Initial programme funded by ONR but major development by BAE Systems Development through laboratory, full scale demonstrators and trials Machinery Vibration Typically tonal

14 Active Vibration Control for Marine Application 40 Mount Resonance Structural Resonances 0 Amplitude (db) x(t) Mode at Hz Y location m X location m k c ,000 Y location m -4-6 Mode at Hz Frequency X location (Hz) m Y location m Fully Active control of Rigid-Body Modes only Mode at Hz X location m 2 4 6

15 Active Vibration Control for Marine Application Main themes of Presentation Machinery Raft Vibration Isolation Magnetic Levitation fully active Smart Spring passive/active hybrid Vibro-acoustic Control Selective Damping Remote Selective Damping Subject of several BAE Systems Patents K

16 Active Noise Control System for Propeller Aircraft Controller with 46 structural actuators and 72 microphones built by Centralised Ultra Electronics digital system and now made fitted by Ultra to over Electronics 1,000 aircraft controls 5 harmonics with 48 structural actuators at 72 acoustic sensors, distributed throughout cabin.

17 Potential Future Open Rotor Aircraft

18 Airbus A400M Aircraft

19 Active control on a much larger aircraft 1kg reduction in weight saves $3,000 in fuel per year

20 Number of actuators required

21 GLOBAL VIBRATION CONTROL THROUGH LOCAL FEEDBACK Centralised feedback Local feedback

22 Simulations with an array of 16 local feedback controllers with force actuators and velocity sensors Random excitation Global vibration response (kinetic energy)

23 Sound radiated by plate as the velocity feedback gain is increased

24 The feedback gain could also be adjusted to maximise the of power absorbed For broadband excitation minimising power requires very nearly the same gain as minimising kinetic energy.

25 Self-tuning to maximise power absorbed The velocity signal can be used for both feedback, using the instantaneous value, and self-tuning, using the mean square value. f v 2 v Power absorbed fv v 2

26 Implications for modular controllers If the actuators and sensors are collocated and dual, an array of independently acting modules with local feedback loops are guaranteed to be stable (Balas, 1979), regardless of 1) changes in the dynamics of the system under control 2) failures in individual modules 3) The positions of the local modules Self-tuning of the feedback gains also allows the performance to be optimised to the environment

27 Motivation for Cochlear Modelling To develop signal processing systems that mimic the healthy cochlea for hearing aids and cochlear implants

28 A simple model of the ear The dynamics of the basilar membrane separating the fluid chambers within the cochlea are modelled as an array of mass-spring-damper systems, each tuned to its own characteristic frequency (Helmholtz 1863, von Békésy 1947).

29 Predicted passive cochlear frequency response The displacement at one place on the basilar membrane varies with excitation frequency to give a frequency response. The frequency response of the passive model is nowhere near as sharp as that observed at low levels in a living cochlea.

30 Measured cochlear motion Level of basilar membrane motion as a function of frequency in response to pure tone pressures at different levels, as measured using the Mossbauer technique by Sellick et al (1982)

31 Source of the cochlear amplifier Note that piezoelectric constant of the motor protein (prestin) is about the same as PVDF

32 Lumped model of the cochlea amplifier Inner hair cells behave as sensors. Each outer hair cells behaves as a sensor, an actuators and a local feedback loop.

33 A coupled model of the active cochlea An array of elements, each with an internal feedback loop, is coupled together through the cochlear fluids

34 Predicted frequency response of the active basilar membrane

35 Measured cochlear motion Level of basilar membrane motion as a function of frequency in response to pure tone pressures at different levels, as measured using the Mossbauer technique by Sellick et al (1982)

36 A state space model of cochlear mechanics x t Ax t Bq t 1. When the cochlear model is linear, e.g. for low levels, its stability and frequency response is predicted from the positions of its poles, which are the eigenvalues of the matrix A. 2. As the cochlear model becomes nonlinear, so that A depends on x(t), its time domain response can be calculated using standard solvers, even if it oscillates.

37 Otoacoustic emissions The combination of the cochlear amplifier and nonlinearity gives rise to various otoacoustic emissions in the ear canal. These are used to diagnose hearing disorders and may help us to understand the inner workings of the ear. Sound in Sound in Sound out Sound out Otoacoustic emissions emissions Basilar membrane motion

38 Time Domain Simulations of Cochlea with Uniform Cochlear Amplifier Gain (stable case)

39 Simulations with Small Random Inhomogeneities in Cochlea Gain (Stable)

40 Simulations with Large Random Inhomogeneities in Cochlea Gain (Unstable)

41 Stability of the Real Cochlea Most people s ears are not stable! In very quiet environments, with no excitation, clear acoustic signals can be recorded in the outer ear in 60% of men and 80% of women, which are called spontaneous otoacoustic emissions: The spectrum of the earcanal signal from a normal human ear (Adapted from Martin et al., 1990a) These emissions are almost pure tones, at frequencies which are stable, but vary from person to person. It would seem that for very low excitation levels, the feedback gains in the cochlea are somehow automatically set very close to the point of instability, which is also the point of maximum sensitivity.

42 ISVR Research Teaching Enterprise