Instrumentation

Automated Analysis and Understanding of Biological Cells from Microscopic Images

Cell identification and counting are experimental procedures used across many scientific disciplines, because of their fundamental-biology and clinical relevance. For example, research in follicle development in mammalian ovaries requires detection, size-measurements and counting of follicles. These techniques are routinely performed manually by trained personnel and are labour intensive. Automation in follicle detection and more generally in cell detection, is, for this reason, very much desirable.

The aim of the project is to develop software capable of processing microscopic images of biological samples and of enabling the user to identify efficiently and count automatically cells. Image processing computer algorithms which are able to identify biological structures on microscope images are already available. Nevertheless, these algorithms require further development to improve their detection rate and to increase their range of use. The existing expertise in pattern recognition and image processing available within the School of Engineering at the University of Kent will provide the required know-how to achieve these tasks. More specifically we intend to use new methods which take advantage of recent developments in pattern recognition efficiency by the introduction of “fusion of multiple classifiers” which overcome difficulties of traditional classification methodologies by combining different pattern recognition algorithms.

Existing collaboration with the Schools of Biosciences and Physical Sciences, University of Kent will provide further expertise and resources for the project.

should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Engineering, Physics or a related discipline. Candidates with a degree in Biology and quantitative skills will be considered. An appropriate degree at Masters level will be an advantage.

Please contact Dr. Luca Marcelli (g.marcelli@kent.ac.uk) for more details.

 

CO2 Flow Metering under CCS Conditions Using Coriolis Mass Flowmeters and Soft Computing Techniques

Measurement of CO2 flow in the CCS chain is also a challenging and pressing problem with the recent development and deployment of the CCS scheme in many countries. CCS is considered as an effective technology to reduce CO2 emission from electrical power generation and other industrial processes. Accurate measurement of captured CO2 is necessary not only for environmental purposes to detect CO2 leakage, but also for verification of the CO2 account under emissions trading schemes. However, CO2 flow in the CCS chain is more difficult to measure than other multiphase flows as the physical properties and flow regimes of CO2 could change significantly through the long-distance transportation, especially with the variations of environmental temperature and pressure or in the presence of inevitable impurities. This project aims to apply Coriolis flowmeters incorporating soft-computing techniques to determine both flowrate and volume fraction of CO2 flow. Hybrid data-driven models and deep learning models will be developed based on experimental data. Meanwhile, the dynamic characteristics of CO2 flow under CCS conditions will be analysed.

Candidates should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr Lijuan Wang (L.Wang@kent.ac.uk) for more details.

 

Combustion Stability Monitoring through Deep Learning

Various techniques are available for measuring flame parameters such as temperature, oscillation frequency etc. in combustion systems. The quantitative assessment of the flame quality (e.g., stability) can then be carried out based on the multiple parameters measured. However, there are still significant errors in the existing techniques due to the complex nature of the combustion process (e.g., multiple parameters). Deep learning technique is a powerful tool to identify the combustion instability and this would significantly advance the state-of-the-art in combustion diagnosis. This project aims to perform an intelligent combustion instability monitoring with the aid of digital imaging, image processing data analytics, and deep learning techniques. The objectives are to develop a prototype imaging system for acquiring flame videos under different combustion conditions; to develop a deep convolutional neural network (CNN) architecture and image processing techniques to process the flame videos; to develop a graphical user interface for real-time prediction of flame instability.

Candidates should have or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr Md Moinul Hossain (m.hossain@kent.ac.uk) for more details.

 

Condition Monitoring of Wind Turbines through Multispectral Imaging

Onshore and offshore wind farms are becoming an increasingly important source of renewable energy and are used in many countries including the UK as part of a strategy to reduce their reliance on fossil fuels. As wind turbines are usually deployed in a harsh environment, it is necessary to monitor the operating state and structural health in an on-line continuous manner. Condition monitoring is an effective way to detect early faults and minimise unplanned maintenance and downtime. However, the current condition monitoring systems mostly focus on the detection of certain drivetrain components and lack of the monitoring of support structures of wind turbines. These single monitoring technique based systems usually have limited reliability and accuracy. In this case, a multifunctional and high-performance system is highly desirable. This project aims to develop a novel measurement system for monitoring the rotor blades and support structure of wind turbines using multispectral cameras. Multispectral image processing algorithms will be developed to detect the defect, deformation and vibration of rotor blades, tower and foundation of wind turbines. Meanwhile, thermoelasticity stress analysis will be applied to detect the internal defect and high stress region. An extensive experimental programme will be conducted to evaluate the performance of the system.

Candidates should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr Lijuan Wang (L.Wang@kent.ac.uk) for more details.

 

Contactless Temperature Measurement of Stored Biomass

As a renewable energy source biomass is now widely used in electrical power generation. Biomass fuels such as wood pellets, olive pellets, straw are often stored in bulk volumes  at biomass power plants. Since biomass differs significantly from coal in terms of size, shape and other physical properties, biomass power plants have experienced a range of new technical challenges. The naturally low ignition temperature of biomass has increased the risk of fire at biomass power plants, so the temperature profiling of stored biomass has become essential for safety reasons. Power station operators have used conventional devices such as a matrix of thermocouples to gain the crude profile of a biomass site, but such devices are intrusive and often moved away from their original location or even ripped out by the biomass due to excessive forces exerted on them. A contactless, three-dimensional, cost-effective temperature profiling system is thus desirable. Acoustic sensors coupled with advanced signal processing and image reconstruction algorithms have been identified as a potential approach to resolving the measurement problem. This project aims to develop a demonstration  system for the on-line temperature profiling of a biomass storage site. The challenges include investigations into the transmission characteristics of acoustic waves in biomass, reconstruction of the temperature profiles from noisy acoustic data, and design and implementation of a cost-effective system suitable for the installation and operation on a large scale (typically 50x25x5m) biomass pellet pile. An extensive experimental programme will be conducted initially on a biomass storage model in a laboratory environment and then on a biomass power plant.

Candidates should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Professor Yong Yan (y.yan@kent.ac.uk) for more details.

 

Decentralised Fault Reconstruction and Fault Tolerant Control for Nonlinear Interconnected Systems

With the increasing requirements for system performance and reliability from real engineering, the study on nonlinear interconnected system becomes more and more imperative. A large-scale interconnected system is prone to suffering from faults due to its wide distribution in space and the interconnections among subsystems. The whole interconnected system may stop working normally even if one subsystem possesses a fault which is not detected in time. Therefore, it is very important to design a robust fault reconstruction and control scheme such that the controlled system has satisfactory performance even if one or more faults/failures occur. This project is concerned with fault reconstruction and fault tolerant control design using only local information. Such a decentralised scheme is convenient for real implementation and can enhance the reliability. The research will focus on the development of novel, theoretically rigorous methods. The developed results will be tested through multi-machine power systems and the other industrial systems.

Applicants should have a good Honours degree (either first class or upper second class) in Engineering, Mathematics or Physical Science or a relevant discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr. Xing-gang Yan (x.yan@kent.ac.uk) for more details.

 

Digital Imaging Based Characterisation of Biomass Particles in Power Generation

As a renewable energy source biomass is now widely used in electrical power generation. Biomass originates from a range of different sources in a wide variety of forms from untreated biomass (straw, pail kernels, etc) to treated biomass (wood pellets, olive pellets etc) and from cultivated energy crops (miscanthus, willow, etc) to residues and waste-derived fuels. Since biomass differs substantially from coal in terms of size, shape and other physical properties, power plants firing biomass have experienced combustion problems which affect flame stability, emission levels and plant maintenance. It is therefore imperative to quantify the physical characteristics of biomass fuels which will inform the combustion engineers for optimised operation of their plants. Digital imaging has been identified as a cost-effective, non-destructive approach to the requirement. This project aims to develop an imaging based portable system for the quantitative analysis of biomass fuels. The challenges include the design and implementation of an image acquisition system suitable for a diverse range of biomass fuels in terms of size, shape and colour and the development of bespoke computer software for the measurement of characteristic parameters such as particle size and shape distributions and surface properties. An extensive experimental programme will be conducted to evaluate the operability and effectiveness of the system. Power generation organisations such as RWE npower, E.ON and Alstom Power are the sponsors of this project.

Candidates should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Professor Yong Yan (Y.Yan@kent.ac.uk) for more details.

 

Gas-solid Flow Measurement in Fluidized Beds through Multi-modal Sensing and Deep Learning

Gas-solid fluidized beds are widely applied in many industrial processes, such as catalytic cracking of petroleum, polymerization, combustion and biomass gasification. Despite a variety of sensors and instruments based on different sensing principles have been proposed for measuring the fluidization dynamic parameters and monitoring the flow status, they all have their limitations and few are practical in industry. This project aims to develop a novel methodology based on multi-modal sensing and deep learning techniques to measure the solids and bubble dynamics in fluidized beds. Multi-modal sensing unit including electrostatic, acoustic, piezoelectric and pressure sensors will be designed to sense the flow characteristics of the complex gas-solid mixture in fluidized beds. Advanced signal processing algorithms and deep learning models will be developed to derive the solids velocity, concentration, particle size distribution, particle agglomeration, flow regime transition, bubble position, velocity and frequency. A prototype instrument will be developed at the end of the project and tested in laboratory conditions.

Candidates should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr Lijuan Wang (L.Wang@kent.ac.uk) for more details.

 

Measurement of Flame Radical Emissions through Hyperspectral Imaging

In the combustion process, uneven fuel distribution in the burners resulted in poor flame stability and low thermal efficiency. Self-excited combustion instabilities can also be involved due to free radical emissions (e. g., OH*, CH*, CN* and C2*) where the determination of the concentration of emissions is crucial to optimise the combustion operation. Because, the optimise combustion operation is closely linked to furnace safety, combustion efficiency and pollutant emissions (NOx, CO and particulars). In the past, various developments were made to investigate the radical emissions adding different broadband filters into the imaging system such as OH* (304nm). However, Hyperspectral imaging techniques would be an innovative technique that can be used to measure the radical emissions from a range of spectrum bands (UV–VIS/VIS-NIR) simultaneously. The aim of the project is to develop a Hyperspectral imaging technique to investigate the flame radical emissions from a range of spectrum bands simultaneously and also to study the characteristics of the flame and their relationship to emissions. The outcomes of the project will give an in-depth understanding of combustion emissions and also lead to an optimised design and operation of the combustion process. The combining of Hyperspectral imaging with machine learning will permit the automatic classification and characterisation of emissions in high spatial and spectral resolution.

Candidates should have or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr Md Moinul Hossain (m.hossain@kent.ac.uk) for more details.

 

Monitoring and Characterisation of Gas Turbine Flames through Stereoscopic Imaging and Two-colour Techniques

A modern industrial gas turbine is regarded as one of advanced power generation systems and increasingly used in power industry worldwide because of the highest combustion efficiency and lowest pollutant emissions over all combustion power generation systems available today. The gas turbine combustor is a multi-burner where flames appear to be premixed (i.e., blueyish and translucent), highly turbulent and inherently three-dimensional (3-D). The complex nature of the gas turbine flame makes it one of the most difficult objects to be measured, particularly its temperature distribution and oscillation frequency. Digital imaging is one of the most promising approaches amongst all the possible sensing techniques for the monitoring of combustion flames, and has been advanced rapidly as a diagnostic tool for installations on pulverised coal-fired furnaces. This project aims to develop a suitable instrumentation system for on-line, continuous monitoring and characterisation of gas turbine flames. A stereoscopic imaging strategy will be adopted, in which a digital camera and a stereo lens will be used to capture the light of flame at two different viewing angles. The images captured by the camera will be combined to generate stereoscopic images of the flames through dedicated image processing algorithms. Due to the factor that the gas turbine flames are blueyish and translucent, a dedicated optical filtering technique will be employed to capture the radiation intensities of carbon particles (C2) at two adjacent spectral bands within the flame. The information obtained will then be used for the temperature determination based on two-colour pyrometry. A photo detector will also be employed to detect the flame signal over a wide spectral range (from 300nm to 1100nm) at a suitable sampling rate (up to 3000Hz). The signal will be analysed in both time and frequency domains to quantify the dynamic characteristics of the flame, particularly oscillatory frequency and local flame stability. The prototype system, once developed, will be evaluated initially on the gas turbine test rig which is to be built in the university research lab and then on an industrial-scale gas turbine. The technique developed can also be used to determine the gas phase temperature of a sooting flame in a coal-fired combustor so that the temperature difference between the gas phase and solid phase can be examined.

Candidates should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr. Gary Lu (G.Lu@kent.ac.uk) for more details.

 

Monitoring and Characterisation of Large Scale Industrial Fires

Monitoring and characterisation of large-scale pool fires have become increasingly important in the field of fire safety and prevention. Quantified information about large-scale pool fires will inform fire engineers to study large scale fire dynamics, develop better fire sprinkler systems, and validate numerical models of industrial fires. With the advent of digital imaging and image processing techniques, vision based monitoring and characterization techniques have the potential to be deployed. The fire imaging system should be capable of measuring a range of fire parameters and suitable for installation on a large scale pool fire test bed in a university or industrial laboratory. Correlations between the measurements and computation modelling results will be established.

Candidates should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Professor Yong Yan (Y.Yan@kent.ac.uk) for more details.

 

Monitoring of Particulate Emissions through Digital Imaging and Light Scattering

It is estimated that the life expectancy of every individual person in the UK is reduced by 7-8 months due to particulate matters in the air with subsequent health costs of £20 billion each year. A range of industrial processes, particularly, those in the combustion, metal, mineral, and chemical industries release more particulates into the atmosphere than other processes. It is therefore imperative to measure accurately the amount of particulate emissions from such industrial processes. This project aims to develop a novel, cost-effective technology capable of monitoring particulate emissions from industrial stacks by combining digital imaging and light scattering techniques. The project has two primary objectives: (1) to develop a demonstration system that can monitor the emissions of particles of variable size distributions from sub-microns to over 100 microns under dry, wet and low dust emission conditions; (2) to study the fundamental characteristics of particulates from typical industrial stacks using the developed technology. The light scattering technique will be used to measure the density of particles smaller than 10 microns whilst the digital imaging technique will be deployed for the measurement of dust density and size distribution of particles greater than 1 micron.

Candidates should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Professor Yong Yan (Y.Yan@kent.ac.uk) for more details.

 

Multi-Agent Model to Simulate Economic and Financial Systems’ Behaviour

The use of interacting agents to model economic markets from the bottom up is becoming a popular research methodology and it is shaping the way we envisage economic models. Agents can be characterised by behaviours which can follow simple rules. Despite the simplicity of the model’s rules, interesting patterns and behaviours can emerge due to the agent interactions.

This PhD project will aim to investigate these complex patterns emerging from financial and economic systems using theory or empirical agent-based models.

Possible topics of research:

a) Following the work proposed by Lux and Marchesi (Nature, 397, p. 498, 1999), the project consists in using a multi-agent model to simulate traders’ behaviour in an artificial financial market. The aim of the project is to characterise how the patterns of behaviour change as the interaction rules change.

b) In their work, (Nature, 469, p. 352, 2011), Haldane and May illustrate how ecology networks can be used to simulate bank failure and subsequent shocks. The aim of this research topic is to develop a computer code, resembling ecology networks, able to simulate simple bank activities. The code will be able to simulate bank failure and how failure spreads to other banks. These phenomena will be analysed as function of number of nodes (banks) and type of connectivity.

Candidates should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Engineering, Computer Science, Quantitative Finance or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr. Luca Marcelli (g.marcelli@kent.ac.uk) for more details.

 

Object Detection through Complex Media (fog/smoke)

Ability to extract visual cues from the surroundings is crucial in most human activities. There are, however, many environmental or man-made scenarios (e.g., fog, cloud, smoke, dust, haze, etc.) where this capability can be severely compromised. It is crucial to develop an intelligent and effective technique to detect an object through poor environmental condition such as fog/smoke for safer transportation systems (self-driving cars, for instance). Existing solutions are mostly based on radio waves. However, there are a number of challenges that need to be overcome such as low resolution and poor optical contrast as well as low signal to noise ratio. This project aims at delivering a technological solution to vision in the fog/smoke through a multispectral imaging technique. The objectives are to design and set up a multi-spectral imaging system to collect data from multiple viewing angles and spectrum under various fog/smoke conditions (construction of a small scalable experimental rig may be necessary); to carry out a systematic experiment generating different dense fogs/smoke and objects of different geometry; to develop algorithms to detect the object location; to generate ground truth that will be essential for the evaluation of the object detection algorithms.

Candidates should have or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr Md Moinul Hossain (m.hossain@kent.ac.uk) for more details.

 

Observer Design for Time-Delay Systems

Systems involving time-delay are of industrial and applications significance and largely fall into two main categories. The first category arises because of the need to model systems more accurately given increasing performance expectations. Many processes, such as manufacturing processes and the internal combustion engine, include such after effect phenomena in their inner dynamics and time delay is also produced via the actuators, sensors and field networks involved in the practical implementation of feedback control strategies. The second class of problems arises when time delays are used as a modelling tool to simplify some infinite dimensional systems, described by partial differential equations. This tool is used for constructing models of distributed systems modelled by partial differential equations where a set of finite dimensional state variables with appropriate time delay characteristics can be used to represent heat exchange processes, for example. Observers are used to estimate unmeasurable states as well as perform condition monitoring and fault detection and isolation tasks. This project will consider the development of robust observers appropriate for time-delay systems.

Applicants should have a good honours degree (either first class or upper second class) or hold an MSc in a relevant Engineering, Mathematics or Physical Science discipline.

Please contact Dr Xinggang Yan (X.Yan@kent.ac.uk) for more details.

 

On-line Monitoring of Particle Size Distribution of Biomass Particles in Power Generation Using Acoustic Sensors

As a renewable energy source biomass is now widely used in electrical power generation. Biomass originates from a range of different sources in a wide variety of forms from untreated biomass (straw, pail kernels, etc) to treated biomass (wood pellets, olive pellets etc) and from cultivated energy crops (miscanthus, willow, etc) to residues and waste-derived fuels. Since biomass differs substantially from coal in terms of size, shape and other physical properties, biomass power plants have experienced a range of problems. For example, wood pellets are a common biomass fuel, which often contains fine wood-dust presenting an explosion hazard. A mechanical separation process can separate fine wood-dust from the wood pellets and then feed directly to the burners. However, the presence of a small proportion of wood pellets in the wood-dust creates combustion problems. It is thus desirable to measure on-line continuously the size distribution of biomass particles in fuel handling pipelines. Acoustic sensors coupled with advanced signal processing algorithms have been identified as a cost-effective, non-destructive approach to the measurement problem. This project aims to develop a novel measurement system for the on-line sizing of biomass fuels. The challenges include the design and implementation of a cost-effective system suitable for a diverse range of biomass fuels. An extensive experimental programme will be conducted at a biomass power plant (e.g. Tilbury Power Station) to evaluate the operability and effectiveness of the system.

Candidates should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Professor Yong Yan (y.yan@kent.ac.uk) for more details.

 

Reaction-Diffusion Model to Study of Morphogen Gradient Formation in Drosophila Embryo

Biological cells use signalling to regulate processes such as recognition, proliferation, differentiation, and apotosis. A common signalling mechanism used by cells involves producing soluble molecules, which diffuse and bind to the inner or outer cell surface. This can be modelled with point particles diffusing in a 3 dimensional medium containing structures, which act as reactive surfaces. This is what happens in cell differentiation which gives rise to spatial pattern formation in organisms. In this process, signalling molecules, morphogens, spread and form a concentration gradient which, once ‘read’ by cells, drives gene expression and ultimately pattern formation. While it is generally accepted that these concentration gradients arise through diffusion, recent studies into the Drosophila (fruit-fly) embryo showed that the mechanisms that drive their formation need better understanding.

Gregor and co-workers [Gregor et al, 2007a] used transgenic Drosophila embryos where the Bicoid protein, an endogenous morphogen, was replaced with a green fluorescent fusion protein (Bcd-GFP) and performed real-time measurements of the Bicoid concentration to characterise its spatio-temporal dynamics. These authors showed the inadequacy of the traditional model, which predicts a time for the formation of gradient which is too long compared to the actual one. The authors proposed the hypothesis that the nuclei within the embryo influence the formation of the gradient when absorbing the Bicoid proteins.

The aim of this research project is to develop a computer code to model the reaction-diffusion within the Drosophila embryo, and test hypotheses regarding the mechanisms underlying the dynamics of morphogen gradients. A reaction-diffusion model, with appropriate boundary conditions to account for the geometrical details of the Drosophila embryo, will be developed to test new hypotheses against the experimental data available in the literature. This work would have an important impact in developmental biology and would help the understanding of cell signalling mechanisms in other biological problems.

Candidates should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Physics, Physical Chemistry, Engineering, Computer Science or a related discipline. Candidates with a degree in Biology and quantitative skills will be considered. An appropriate degree at Masters level will be an advantage.

Please contact Dr. Luca Marcelli (g.marcelli@kent.ac.uk) for more details.

 

Reconstruction of Radical Emissions in a Combustion Flame through Stereoscopic Tomography

The changes in fuel supplies in the coal fired power generation industry, particularly where biomass is used, have been posing significant technical challenges for combustion plant operators and engineers to maintain high combustion efficiency and low atmospheric emissions. A flame is regarded as the primary reaction zone of the highly exothermic reactions of fuels and contains important information relating closely to the energy conversion, pollutant formation process and thus the quality of the combustion process. This project aims to develop a stereoscopic imaging system for the reconstruction of the radical emissions within a combustion flame. The primary objectives of the project are: 1) to design and implement a prototype imaging system capable of reconstructing the radiative characteristics of the free radicals (e.g., OH*, CH*, CN* and C2) based on stereoscopic tomography; and 2) to establish the relationships between the radical characteristics obtained by the imaging system and the emission levels in flue gas based on soft computing techniques, such as neural network, fuzzy logic and data mining. The novelty of the proposed new system is primarily that by applying the stereoscopic tomographic technique, the three-dimensional reconstruction of flame radical emissions can be achived by a single camera. The prototype system, once developed, will be tested on a gas-fired combustion test rig in the University Instrumentation Lab. The results will be compared with that obtained by a spectrometer to assess the accuracy and reliability of the system. The relationships between the emissive characteristics of radicals and the chemical/physical properties of the fuels, together with the flue gas emissions (e.g., O2, CO, NOx and SO2 which can be obtained using a gas analyser) will then be established. The outcome of this project will help to predict emissions directly from the flame information instead of the flue gas measurement, shortening the control loop for emissions reduction. RWE npower (UK) is the sponsor of this project.

Candidates should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr. Gary Lu (G.Lu@kent.ac.uk) for more details.

 

Robust Decentralised Output Feedback Control for Complex Interconnected Systems

With the advancement of scientific technology, the dynamical systems used to model reality have become more and more complicated. The complexity mainly lies in nonlinearities, uncertainties, strong interconnections among subsystems and high dimensionality. Such systems are often modelled as dynamic equations composed of interconnections of a set of lower-dimensional subsystems, and are termed large-scale interconnected systems. How to deal with the interconnections in the control design to reduce the conservatism is full of challenge specifically when the interconnections are fully nonlinear. This project will consider decentralised output feedback control design for complex interconnected systems using the structure of the isolated subsystems and the interconnections. The research will focus on the development of novel, theoretically rigorous methods for decentralised control. The developed results will be tested through industrial systems.

Applicants should have a good Honours degree (either first class or upper second class) in Engineering, Mathematics or Physical Science or a relevant discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr. Xing-gang Yan (x.yan@kent.ac.uk) for more details.

 

States and Parameters Estimation for Complex Networked Systems with Applications

With the advancement of scientific technology and the increasing requirement for high levels of system performance, the dynamical systems used to model reality in demanding such as power networks, ecological systems, biological systems and energy systems, have become more and more complex. It is clear that for practical systems, states and parameters estimation is absolutely important to guarantee the control scheme’s implementation or to replace high cost sensors.

It is well known that structural constraint has been widely employed in observer, which has greatly limited the application of the estimation theory. This project is concerned with development of novel, theoretically rigorous methods for states and simultaneously parameters estimation based on system structure characteristics. It is expected that this research yields high quality research work by developing novel conditions to relax the traditional structural constraints from theory point of view such that the developed results can be applied to a wide class of systems. Both the advantages of adaptive techniques and the characteristics of variable structure systems will be combined together to reduce the conservatism and enhance the robustness. The developed results will be tested by real systems such as automobile suspension system, robot system or some industrial systems.

Applicants should have a good Honours degree (either first class or upper second class) in Engineering, Mathematics or Physical Science or a relevant discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr. Xinggang Yan (x.yan@kent.ac.uk) for more details.

 

Three-dimensional Emission Tomography of Flame Chemiluminescence

The process tomography is a very useful technique for three-dimensional (3-D) reconstruction and characterisation of many industrial processes including combustion process. The emission tomography of chemiluminescence (CTC) is a 3-D imaging technique that can resolve the internal structure of a flame in a combustion system so as to reveal spatiotemporally complex combustion phenomena in such as system. Various techniques were developed in the past to measure flame chemiluminescence emissions, but there are still many technical problems (such as complex system setup, low temporal/spatial resolution, and poor measurement accuracy) which need to be resolved. This project aims to examine a 3-D imaging technique which works based on flame chemiluminescence detection. The main objectives are to study and implement an image-fibre based high-speed tomographic imaging system for acquiring flame images; to reconstruct flame chemiluminescence emissions using acquired images to study optical tomographic and digital image processing techniques; to structure a simple graphical user interface for operating the system and presenting the flame images/reconstruction; to conduct experiments in lab-scale combustion rig and to evaluate the methodology developed.

Candidates should have or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr Md Moinul Hossain (m.hossain@kent.ac.uk) for more details.

 

Tomographic Reconstruction of Soot Volume Fraction and Emissivity of a Sooting Flame using Multiple Cameras

Within the scope of the advanced flame monitoring techniques, reliable and continuous measurement of temperature and soot volume fraction (i.e., concentration) of a sooting flame (such as a pulverised coal fired flame) plays an important role in the in-depth understanding of fuel combustion and pollutants formation processes. However, due to the uncertainty of the flame emissivity, particularly where fuel blends (mixture of different coals, or coal and biomass) are used, the measurement of the flame temperature and its distribution presents the most technical challenge to researchers and engineers. In practice, a flame has to be assumed to be a grey-body (i.e., the emissivity is constant) and thus the multi-colour pyrometry can be applied for the temperature determination. This would introduce a un-neglectable error in the measurement results. In addition, very limited work was previously conducted on the online measurement of soot volume fraction in a coal-fired flame. A research project is therefore proposed to develop an instrumentation system capable of the three-dimensional monitoring of the flame soot volume fraction, emissivity and their fluctuations based on optical tomographic and two-colour techniques. In the project, an optical transmission system will be designed to capture the light of flame from three different directions and transmit the images onto three identical RGB CCD cameras. Computing algorithms will be developed for reconstructing the soot concentration, emissivity and temperature cross flame sections using the optical tomography and two-colour principle (based on the Planck’s radiation law). Quantitative relationships between soot volume fraction, emissivity and the temperature, together with fuel/air inputs and emission levels, will be established. An extensive experimental programme will be conducted on both laboratory and industrial-scale combustion test facilities to evaluate the operability and effectiveness of the system.

Candidates should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr. Gary Lu (G.Lu@kent.ac.uk) for more details.

 

Vibration Measurement of Rotor-bearing Systems Using Electrostatic Sensors

Rotating machinery such as generators, electromotors, compressors and wind turbines accounts for a large proportion of mechanical devices in the electrical, energy, chemical, aviation and other industries. The rotor-bearing systems in these devices usually suffer from different levels of damage and faults due to physical or chemical changes in the process of long-term operation, resulting in system downtime, economic loss or even disastrous consequences. To reduce system downtime or avoid unnecessary damage of mechanical devices, a condition monitoring system should be applied to obtain operational parameters, identify health status and establish maintenance strategies. This project aims to develop a vibration measurement system for rotor-bearing systems using electrostatic sensors. The challenges include the design of electrostatic sensing unit and development of signal processing algorithms for the measurement of vibration parameters. An extensive experimental programme will be conducted to evaluate the effectiveness of the system.

Candidates should have, or are expecting to obtain in the near future, a First Class or good 2.1 Honours Degree in Electronic Engineering, Computer Science or a related discipline. An appropriate degree at Masters level will be an advantage.

Please contact Dr Lijuan Wang (L.Wang@kent.ac.uk) for more details.