A
Proposal Submitted to the
Transport
Asset Management Competition
Artificial-Intelligence-Based
Optimization of the Management of Snow Removal Assets and Resources
by
Md.
Salim, PI
Marc A. Timmerman, Co-PI
Tim Strauss, Co-PI
Michael E. Emch, Co-PI
1
Sabin Hall, Mail Code 0406
Respectfully
submitted
for the MTC's consideration on
April
3, 2000.
0.0 Abstract
A knowledge-base will be implemented using
existing GIS software and an artificial intelligence shell to optimally manage
snow removal assets. The knowledge-base will be fully interactive and include
provisions for entering meteorological observations and field data to refine
the snow removal plan. The knowledge-base will also be capable of doing what-if
studies on construction projects still in the planning stages for the purpose of
evaluating the impact of the construction on snow removal. The project also
includes an interactive website, an Advisory Committee, a pilot project of
actual snow removal plans for a mid-size municipal area, and other educational
and publicity enhancements.
1.0 Qualifications of Research Team
Md. Salim, PI, is an associate professor of Construction
Management at UNI with a Civil Engineering background. He is the author of
numerous papers in the areas of economic and artificial intelligence aspects of
transportation resource asset management. These include studies on
transportation systems and their impact on economic development. Some research
projects of relevance to the proposed work include the development of a
computer-based materials inventory and control system for public utility
companies (funded by a utility company), the impact of bridge deck cracking on
durability (funded by the Iowa Department of Transportation), the development
of an artificial intelligence (AI) supported system for rapid prototyping, the
development of an expert system for diagnosing cracks in concrete structures,
the development of an Expert System for troubleshooting in furniture factories,
and the development of a Process Model for construction site operations
Marc A. Timmerman, Co-PI, is an assistant professor of
Electro-Mechanical Systems at UNI with a Mechanical Engineering background. He
is the author of numerous papers and successful grant proposals in the general
area of optimal and intelligent controls. He brings a deep background in the
area of optimization methods in general and numerical artificial-intelligence
optimization techniques in particular. He has numerous grants in the areas of
optimal and intelligent controls funded by NASA and the Space Missile Command
of the Air Force.
Tim Strauss, Co-PI, is an assistant professor of Geography at UNI
with a background in Geography and Economics. He is an expert in the area of
GIS and has had numerous grants and publications in this area. Some
representative grants of relevance to the proposed work include several grants
funded by the Iowa Department of Transportation for the development of the
"Smart Map," an innovative GIS-based tool for identifying the
location of road accidents.
Michael E. Emch, Co-PI, is an assistant professor in Geography at UNI
and has a background in Geography and Biology. He is a specialist in the are of integrating public records databases, such as
public health information, with GIS information. He is the recipient of a
Fulbright Scholarship and is the author of numerous grants and publications.
2. 0 Topic of Research
The proposed research seeks to build a
knowledge-base that would allow a public works department (a) to optimally
manage assets for snow removal and (b) to use what-if simulations to assess the
impact of changes in snow removal assets or of environmental changes like new
construction. The final product of this research will be a user-friendly
data-base oriented software tool. A municipality or other agency responsible
for snow removal would use this tool to maintain a database of snow removal
assets and resources. The municipality would also have a GIS database with an
up-to-date geographical layout of the agency's service area. When a snowfall is
forecasted, the meteorological details of the forecast (time, accumulation,
temperatures) will be entered into the system. The system will run various
scenarios and generate a detailed snow removal plan based on the forecasted
snow-fall. The agency will begin allocating and moving assets and resources
immediately. A major benefit of the systems will be the ability to order needed
materials intelligently before a snowfall when transportation is still possible
and to place snowplowing equipment in the field prior to the snowfall thus
eliminating problems of needing to plow access roads to reach main roads.
During and after the snowfall updated meteorological predictions will be
entered into the system. Direct field measurements and observations of snow
accumulations will also be entered into the system and the snow removal
schedule refined. Finally, after the plowing starts meteorological predictions
and field data can continue to be entered into the system for further on-line
refinements.
A secondary feature of the system will be the
ability to simulate snow removal for facilities still in the approval or
planning stage. The geographical parameters of new roads or constructions are
added to the existing data-base model and simulations are run detailing the
impact of the proposed works on the agency's snow removal infrastructure. This
model will include refinements such as calculations for groundwater runoff for
flood mitigation studies, and salt runoff for environmental impact studies or
for predicting the corrosion damage of roadway infrastructures such as bridges
and road surfaces. This simulation capability will be extremely valuable for
environmental impact reports and for estimating the real maintenance
costs of proposed private and public works construction.
2.1 Importance
Snow removal and the stress of snow removal
materials on public structures are an enormous budgetary burden on
municipalities in cold climates, for example the City of
Researchers have discussed the possible use
of decision-science tools for optimizing asset management of snow removal
resources. This project proposes to combine all of these element into a single knowledge-base that will use GIS
spatial information, asset, resource, and traffic information in existing
databases, and artificial intelligence decision-science optimization tools, to
optimally manage assets for snow removal. Obviously snow removal is an area
of significant expenditure by agencies responsible for maintenance of
transportation infrastructures and the possible costs savings of using better
asset management tools is significant.
2.2 Relevance to MTC
The current Strategic Plan for the Midwest
Transportation Consortium outlines three salient areas of relevance for the
MTC in section 1B:
(1) "The theme
of the MTC is Sustainable Transportation Asset Management and
specifically the application of sustainable asset management principles and
techniques to transportation infrastructure …In the next decade, public and
private transportation organizations will require an increasing number of highly
qualified young professionals with expertise in transportation management
systems and operations." [emphasis added]
(2) "Tomorrow’s
transportation professionals will require basic, advanced, and continuing
education, as well as familiarity with new and nontraditional approaches and
technologies, to effectively operate and manage transportation
systems." [emphasis added]
(3) "To
accomplish this transition, we propose a program centered on the philosophy
that interdisciplinary approaches, blending sound engineering technology
with fiscal and economic analysis, logistics, and new management practices and
communication technologies, would provide tomorrow’s professionals with the
tools to be successful." [emphasis added]
The proposed work greatly furthers all three
of these strategic mission goals of the MTC:
(1) The proposed
work has a high emphasis on education and involves graduate and undergraduate
students directly in the research. The investigators will include themes from
this work in their instructional activities and in their educational
publications. A fully interactive website is part of this proposed work that
will showcase the work to students and future students. The work will be
instrumental in exposing college and pre-college students to careers in
transportation studies.
(2) The proposed
work will extend cutting-edge methods in artificial intelligence to problems in
asset management for transportation. A completely new way of using GIS is
proposed involving an interaction between artificial intelligence optimization
tools and exiting GIS technology. The work will include a pilot study of snow
removal in the
(3) The project team
is highly interdisciplinary including Civil and Mechanical Engineers and
Geographers with extensive experience in artificial intelligence,
controls/optimization, economic analysis, database integration, and GIS. The
proposed work will create an immediately useful software tool for
transportation management professionals.
Is this work sustainable? There are
excellent opportunities for commercialization of this software and a strong
emphasis of the later stages of the work will be in securing a continued
funding for this project from commercial sources.
3.0 Literature Review
The literature in this area is very extensive
and a selection of recent relevant articles is presented. The three main
thrusts of scholarly work in this area concern database integration and
knowledge-base development, asset management uses of GIS, and optimization
theory as applied to transportation problems.
(1) In the area of
knowledge-base development some recent general paper include that of Begur and coworkers (1997) who integrated GIS information
to optimize the routing of visiting nurses and Weigel
and Cao (1999) who optimized the routing of Sears service vans using operations
research methods combined with GIS to save $42 million per year. In
applications of relevance to municipal public-works projects Tsai and Frost
(1999) have integrated GIS databases in a knowledge-base for toxic waste
remediation, Kamler and Beckel
(1999) have described the integration of GIS with other databases for public
transport management, and one municipality has even combined GIS database with
public health data to help combat mosquito infestations (1997). Section 2.1
summarizes recent publications in the application of GIS and database
integration for snow removal problems. These papers stem from a seminal work in
1987 by
(2) In asset
management applications of GIS some relevant work in public works areas
includes that of Chang and coworkers (1997) in municipal solid waste removal, Yeh and Tram (1997) in electric power delivery, and Taher and Labadie (1996) in municipal water distribution.
One of the seminal papers in this area is that of O'Neil (1991) who described
the use of GPS in municipal transportation planning.
(3) The applications
of optimization theory to transportation problems is
extremely extensive as is demonstrated by an enormous body of literature.
Typical problems are optimization of public transit routes, optimal use of
vehicles for public transit, and optimal deign of road improvement for given
traffic patterns. These optimization techniques fall under four major headings;
probability techniques, heuristic techniques, artificial intelligence
techniques, and classical Variational calculus
techniques.
4.0 Research Tasks
Research Methodology: The object of the work is to create a knowledge-base
with the following capabilities:
(1) Input
geographic data from a GIS package.
(2) Overlay traffic
loads from public-works databases.
(3) Overlay snow
accumulation data from field studies.
(4) Overlay
drainage information from public-works databases or studies.
(5) Maintain a
database of all snow removal assets: equipment and de-icing materials including
estimates of efficiency, melting rates, etc…
(6) Establish a
weighted performance index of desired outcomes
The cost
index is numerical representations of all of the above times weighing factors.
(7) Implement an
Artificially Intelligent (AI) solver to OPTIMIZE the solution of the
performance index with the SNOW REMOVAL ASSETS as a fixed constraint.
(8) Generate a
comprehensive strategy to remove the snow.
(9) Input field
data during snow removal and re-calculate the optimal strategy to refine the
strategy.
(10) Allow the
users to play "what if games" like assessing the impact of new
construction on the snow-removal plan. Provide studies useful for zoning and
environmental planning purposes.
This knowledge-base will be implemented using
established programs like the GIS package ARC/INFO and readily available
artificial intelligence shells. No funds are sought for software development, the program consists entirely of software
integration and engineering.
4.1 Project deliverables
The deliverables will be the following:
(1) A fully
integrated working knowledge-base software package capable of:
(a) Inputting GIS
information, snow removal asset information, meteorological predictions and field
data.
(b) Generating
optimal snow removal plans based on this data.
(c) Entering data on
proposed construction and infrastructure changes and simulating snow removal
and environmental impacts of the proposed changes.
(2) Complete
documentation for this knowledge-base package.
(3) A full-scale
case study for the knowledge-case based on the
(4) An interactive
website showcasing this research.
(5) Positive
publicity for the MTC through professional publications and presentations and
showcasing of the project to UNI students and prospective students.
(6) Positive
publicity for the MTC through the projects advisory board and the project's
direct contribution to public works agencies in the
(7) Extensive
involvement of graduate and undergraduate students in this research exposing
students to career opportunities in the transportation studies area.
5.0 Project Timeline
5.1 Duration—24 Months, Summer 2000 to
Spring 2002.
5.2 Graphic Timeline
|
ACTIVITY |
YEAR ONE--SEMESTERS |
YEAR TWO-SEMESTERS |
||||
|
|
SUMMER 2000 |
FALL 2000 |
SPRING 2001 |
SUMMER 2001 |
FALL 2001 |
SPRING 2002 |
|
I. SOFTWARE REVIEW AND EVALUATION |
X |
X |
X |
|
|
|
|
SOFTWARE ACQUISITION |
X |
X |
X |
|
|
|
|
LAB SET-UP |
X |
X |
X |
|
|
|
|
SOFTWARE INTEGRATION |
|
X |
X |
|
|
|
|
II. PRELIMINARY SOFTWARE VERIFICATION |
|
|
X |
X |
|
|
|
GIS ENTRY |
|
|
X |
X |
|
|
|
ASSET ENTRY |
|
|
X |
X |
|
|
|
FIELD DATA ENTRY |
|
|
X |
X |
|
|
|
TEST CASE -OPTIMIZATION |
|
|
|
X |
|
|
|
TEST CASE -VALIDATION |
|
|
|
X |
|
|
|
PRELIMINARY MANUAL |
|
|
|
X |
|
|
|
III. FINAL SOFTWARE VERIFICATION |
|
|
|
|
X |
X |
|
GIS ENTRY |
|
|
|
|
X |
X |
|
ASSET ENTRY |
|
|
|
|
X |
X |
|
FIELD DATA ENTRY |
|
|
|
|
X |
X |
|
REAL CASE - OPTIMIZATION |
|
|
|
|
X |
X |
|
REAL CASE-VALIDATION |
|
|
|
|
X |
X |
|
FINAL MANUAL |
|
|
|
|
X |
X |
|
IV. DISSEMINATION |
X |
X |
X |
X |
X |
X |
|
PUBLICATION AND PRESENTATION |
|
|
X |
X |
X |
X |
|
WEBSITE IMPLEMENTATION |
|
|
X |
X |
X |
X |
|
REVISED MANUAL |
|
|
|
|
X |
X |
|
REVISED SOFTWARE |
|
|
|
|
X |
X |
|
COMMERCIALIZATION |
|
|
|
|
|
X |
5.3 Project Milestones
The project is organized into three
sequential phases (labeled as I, II, and III in the chart) and one concurrent
phase (labeled as IV in the chart). The keys milestones are as follows:
(a) I. Software
Review and Evaluation – ARC/INFO will be used as the GIS package. There are
several possible choices for artificial intelligence shells and in this phase
this phase such shells will be evaluated for speed, cost, and compatibility
with ARC/INFO and with the project goals. By the end of the Spring
2001 semester a lab will be set-up with all the software installed.
(b) II. Preliminary
Software Verification – a simple small size verification problem will stated
with an easily checked solution. The entry of GIS, asset and field data will be
verified. A simple and easily verified, optimization problem will be solved. By
the end of the Summer 2001 term it will be possible to
verify the operation of the package on a small-scale, easily verified test case.
A preliminary version of the manual will also be prepared.
(
c) III. Final Software
Verification – this phase is similar to phase II
except that a test case comprising of snow removal for the
(d) IV.
Dissemination – will occur concurrently with the phases I, II and III, with a
web-site launched early in the program (Summer 2000) and preliminary papers
presented in the Spring of 2001. Towards the end of
the grant (Spring 2002) papers with advanced results will be presented and work
will start on attracting a commercial sponsor for the project.