|
Research Projects
Title: Pipeline Intermodal System to support Control, Expedition and Scheduling
(PISCES)
Funding: EU
Researchers: Abdennour El Rhalibi
Grant Holder: Professor
Gerry Kelleher
Project Manager: Dr. Farath Arshad
The PISCES project will address the difficulties of providing information
through the transport chain to enable the efficient implementation of logistic systems
supporting the movement of goods. PISCES will demonstrate that intermodal transport can be
managed more effectively, is more attractive and can be focused on more environmentally
benign and sustainable modes if systems are in place which enable smooth transfer of
information between the parties involved.
The project is addressing the problem of information provision to the users of the
chains. The current "flow" of consignments through the intermodal transport
chain is effectively only sequence of stages, and at each of these stages the cargo is
halted while it is re-marshaled, re-loaded, and allocated "new" control
references in new systems. PISCES will enable the complete transport chain to be scheduled
and managed as a single process, over a range of modes and with different providers,
without the need for all the participants to adopt the same information system or even the
same information standards. Other projects have sought to address these issues but these
earlier initiatives have sought to have access to vested interest "own"
information to which they are currently not privileged to receive, or they have used IT
which is intrusive or costly to many of the participants. In the worst cases projects have
required all parties to adopt a proprietary additional system or application. For all
these reasons very limited success has been achieved to date. This has lead to shipping
lines, in particular, trying to impose their own bespoke system on their clients - an
approach which is resisted by the clients as it imposes extra costs on them and tends to
limit the range of shipping choices available (this is a major problem for SME's trying to
interact and exploit intermodal transport systems). The aim is to provide an information
transparent transport system that allows all members of the transport chain to know
exactly what is happening to freight, when, why and where. This will allow the supplier to
maintain better control of their goods transfers.
As an essentially practical project PISCES is heavily biased towards the practical
realisation of solutions to these issues and the provision of demonstrations of the
technology developed. The general issue of information transfer within intermodal
transport chains is too big to be tackled within one project with any degree of certainty.
The project thus concentrates on an important subset of these chains (through and either
side of the port) and contains partners with expertise in all elements of that chain and
in the provision of information systems to support the chain. The project is user driven
and ensuring a project which will concentrate on the solution of the real problems
encountered by the industry. The knowledge and understanding of the transportation
Industry by the participants is also demonstrated by the belief that costs of provision
and operation of this service, must be recouped through real operational savings by the
services providers, which in turn is passed on in benefits and savings to the end users.
For this reason the project is focused on the exchange of operational critical information
in real time, at the time it is needed. PISCES is therefore different from earlier
projects which have had a documentary focus , seeking savings only through reduced manual
effort. The demand for this development is already evidenced by the investment many large
end users have made in their own domestic control systems. Competitors within USA and Asia
are already investigating the provision of these systems funded by major corporations and
government.
Partners
Fraser Williams Logistics (UK)
Title: Combined Adaptive and Robust Methods for Real-Time Control of Mechanical
Linkage Systems
Funding: EPSRC
Researchers: Mr. Z. Yuan
Supervisor: Dr. Hongnian Yu and Dr. M. Gilmartin
The overall goal of this research is the development of a robust combined
control theory of nonlinear systems with specific reference to the control of mechanical
systems. The project should provide knowledge (design guidelines related to integrated
design of control systems and mechanical systems) about how to improve the design of the
mechanism so that the overall performance is increased or the same performance is obtained
at a lower cost. The project is a collaboration between researchers in Advanced Control
Engineering and Mechanical Engineering.
The main aim of the project is to improve the current control methodologies for
mechanical systems. The research comprises three overlapping stages:
(a) development of mathematical models of mechanical systems,
(b) a theoretical investigation of the different control philosophies for dealing with
uncertainties when trajectory tracking, The details for each stage are as follows:
(a) The modelling part consists of:
1) Modelling and identification of the mechanical system with friction and backlash in
mind;
2) Modelling of electrical actuators and sensors;
3) Integration of 1) and 2);
4) Simulation and validation of the mechanical system.
(b) Theoretical investigation of the different control philosophies
The problem of position control in the presence of uncertainties will be considered. A
number of different control schemes, including stability based adaptive control and robust
variable structure control will be studied in computer simulations and compared on the
basis of stability, tracking performance, and computational efficiency for real-time
implementation. The main theoretical work will focus on combining the different control
philosophies, using Lyapunov stability theory; the combined controller will possess the
positive features of adaptive and variable structure control philosophies. The research
will investigate the degree of robustness and the degree of stability for bounded
disturbances, of the combined algorithm. Finally the combined algorithm will be optimised
from a computational view point, and tested in computer simulations. The tests will
consider the models developed in (a) and will simulate a range of structured and
unstructured uncertainties. The dynamic models used for the simulations will also consider
issues such as friction (which may take upto 25% of drive torque) and actuator dynamics.
(c) Implementation and experimental validation of the control methods
The combined robust and adaptive control method will be implemented on the mechanical
system developed at JMU. The test program will closely mirror the computer simulation
tests. During the experiments, the payload will be changed to test for structure
uncertainty and disturbance added at both the inputs and outputs channel to test the
unstructured uncertainty.
Title: Time-critical Rescheduling Using Truth-maintenance (TRUTH)
Funding: EU
Researchers: Dr. Jon Spragg
Supervisor: Professor Gerry Kelleher
Partners: British Maritime Technology (UK), Iberia Airlines (Spain), Pirelli
(Italy), SISCOG (Portugal), SYSECA (France), SI, Center for Industrial Research (Norway),
Liverpool John Moores University (UK)
The TRUTH project is funded by the European Community under the Esprit program.
The partners on the project are based in the UK, Italy, Spain, Portugal, France and Norway
and include both research centres and some of the largest commercial/industrial companies
in Europe.
TRUTH addresses the creation of rapidly reconfigurable systems capable of scheduling
and reactive re-scheduling. The solution of complex scheduling problems of large size is a
computer- intensive and usually time-consuming activity, of trying to find a solution
which satisfies a great number of constraints whilst optimising on a multiplicity of
criteria. The output of all this automatic compiler work tends therefore to be obscure.
Scheduling systems are thus normally poor in self- explanatory power and deficient in
their ability to deal with change - that is they are inflexible and cumbersome when one
wishes to alter a schedule to accommodate change. Unfortunately change is a fact of life
in most real world contexts (machines break down, staff become unavailable or new high
priority jobs arrive) and it is just such responsiveness to change that is required.
It is expected that the utilisation of the system will lead to cuts in the
reaction-time to changes, a drop in stocks, a rise in turnover and utilisation rate and an
increase in profit margins. Due to the modular design, TRUTH will be cost-effective for
small as well as large applications. The proposal comes at a time when manufacturers have
to pay careful attention to the JIT policies of their customers and will help redress the
balance that presently favours the Far-East in this area. TRUTH has a clearly defined
policy towards exploitation and dissemination of results which will bring home to
manufacturers and transport industries the advantages and benefits of its adoption. The
exploitation policy will be based on one hand on the wider utilisation by the end-users of
the results obtained and on the other on the commercialisation of the generic results by
the software houses involved in TRUTH.
Title: Controllable Multimedia Presentation Systems
Funding: JISC/ESRC
Researchers: Dr. Farath Arshad
Supervisor: Professor Gerry Kelleher
This project is concerned with developing a controllable, responsive, tool for
the construction and delivery of multimedia materials within an educational and training
context, including an advisory component providing for control in presentation and
progressive adaptation to interaction. It is proposed that techniques for controlling the
presentation of information in an intelligent educational advisory system, SOLA* (Arshad
and Kelleher, 1993) may be used within a general multimedia environment. The essence of
the idea is to use a model, including the requirements of the user, the potential topics
to be presented, their inter-relationships and organisation, in order to control the
presentation of material. In this way, it is possible to make tractable the problem of
what to present to the user, and in what modality and interaction style.
Title: Investigating Formal Models of Change in Artificial Intelligence
Funding: JMU
Researchers: Mr. Aladdin Ayesh
Supervisor: Professor Gerry Kelleher
This project is concerned with domain constraints - establishing possible
solutions for the problems connected with DDC (Directed Domain Constraints). These
problems need to be addressed if any attempts are to be made to extend the theoretical
account of DDC.
The possible extention of the DDC could cover a wider range of internal
relations in the domain. It is intended that the project will determine other possible
types of domain constraints - i.e., the use of domain constraints and their descriptions
of state progression. It is hoped that our approach will yield a general theory covering
both the progression of databases and possible planning states.
Title: Intelligent Tutoring Systems
Funding: JMU
Researchers: Mr. Dhiya Al-Jumeily
Supervisor: Dr. Paul Strickland, Professor G Kelleher, Dr K Whiteley
The aim of this research is to develop a simplified computer algebra system
(CAS), that requires less rigorous mathematical knowledge to be able to operate, and can
allow students to focus on tactical, manipulation skills. A new type of CAS application,
called MathsWeb has been developed. It has the objective of replacing pencil-and-paper
algebraic work with an easy to use intelligent classroom training ground. It has been
implemented for use over the World Wide Web to enable the drill-and-practice of tactical
algebra skills to have an immediate, intelligent feedback, whilst allowing student
performance data to be collected. The purpose of the intelligent feedback is to prompt and
reinforce correct manipulation. This is achieved through the system's ability to produce
symbolic solutions, using a step-by-step approach, depending on (and appropriate to) the
individual user.
The process of this system can be considered as a "white box/black box"
model. A "black box" corresponds to a one-step solution, a "white box"
model corresponds to a full step-by-step solution. Evaluation studies with pupils from a
local secondary school have been carried out within the research, which provides evidence
that in using this model, pupils can develop a full range of algebra skills and more
importantly, step-by-step algebra manipulation.
In order to implement the black box/white box interface, intelligent algorithms have
been developed that identify approach and provide the necessary feedback. These are
explained in detail along with their theoretical proofs.
Title: Design and Evaluation of Hypermedia Learning Material
Funding: JMU
Researchers: Miss. Heather Ainsley
Supervisor: Dr C Ghoaui, Professor G Kelleher, Dr K Whiteley
Hypermedia's support for dynamic linking of chunks of on-line information causes
problems for authoring and usability. Hypermedia documents (or hyperdocuments) lack the
physical attributes of paper based documents that allow us to infer the breadth and scope
of the information contained. Moreover, badly designed hyper-structures impede user's
information retrieval, and can cause 'lost in hyperspace' syndrome. Being a relatively new
phenomenon no standards have yet emerged to guide hypermedia design.
The aim of our research is to develop authoring tools to support structuring of
hypermedia learning material. We are investigating existing design models of learning
material, both paper and computer based, in order to abstract generic structuring
methodologies. A methodology(ies) will then be realised in the form of guidelines or
software tools to automate/semi-automate the design process for authoring hypermedia
learning material. Such guidelines may be suitable for use as benchmark(s) against which
one can evaluate existing hypermedia documents.
Title: Modelling and Scheduling of Manufacturing Systems
Funding: JMU
Researchers: Mr. Antonio Reyes Moro
Supervisor: Dr. Hongnian Yu , Professor G. Kelleher and Dr. K. Whitely
Production scheduling is concerned with the effective allocation of resources
over time. The purpose of scheduling is to determine when to process which job by which
resources so that production constraints are satisfied and production objectives are met.
On-line scheduling is the automatic rescheduling of the system in response to, either
disturbances in the plant operating mode, or to changes in product demand. The objective
of this research is to make a contribution to the solution of the scheduling problem
through the combination of the Petri net and Heuristic based techniques developed by Dr.
Yu and the constraint based reasoning techniques developed by Prof. Kelleher.
In this work the discrete-event scheduling of manufacturing systems is modelled by
Petri nets in a computer simulation and optimal scheduling algorithms are generated using
AI-based heuristic searches (e.g constraint based reasoning methods), which satisfy
specified cost or reward function criteria. An example of a cost function criterion is the
minimisation of the production cycle time or makespan, but other criteria can be chosen
depending upon the system operating goals.
Title: Dynamic Re-scheduling for Resource-constrained Allocation Problems
Funding: JMU
Researchers: Mr. Abdennour El Rhalibi
Supervisor: Professor G. Kelleher and Dr. Hongnian Yu
Aim of this project is to solve a class of real-world resource-constrained
scheduling problem, that exists within the context of dynamic and unpredictable
environments, where the details of the problem are often incomplete, and subject to change
over time. This research work is going to deal with the challenging problems on this area
using complementary approaches namely CSP, Genetic Algorithm and Petri net model.
Title: An Economic Agent Based Approach to Distributed Scheduling
Funding: JMU
Researchers: To be appointed
Supervisor: Professor G.Kelleher
The objectives of the research include exploiting a distributed agent
architecture to support the creation of schedules based on models of economic decision
making. The intention is to provide real improvements in schedule performance by informing
aspects of local schedule decision making with relevant information from related schedules
and exploiting informed approaches to schedule modification.
This project will mutually support, and be supported by, the current running EU
supported project on intermodal transport logistics (PISCES).
|
