Computer Model
Uploaded by
12310-214Computer Model
Uploaded by
12310-214What is a Computer Model?
A computer model is a computer program that attempts to
simulate a real-life system. In other words, it is a ‘virtual’ version of
something in the real-world.
The computer model is designed to behave just like the real-life
system. The more accurate the model, the closer it matches real-
life.
Why Are Computer Models Used?
There are several reasons that computer models are used…
1. To test a system without having to create the system for real
(Building real-life systems can be expensive, and take a long
time)
2. To predict what might happen to a system in the future (An
accurate model allows us to go forward in virtual time to see
what the system will be doing in the future)
3. To train people to use a system without putting them at risk
(Learning to fly an airplane is very difficult and mistake will be
made. In a real plane mistake could be fatal!)
4. To investigate a system in great detail (A model of a system can
be zoomed in/out or rotated. Time can be stopped, rewound,
etc.)
Examples of Computer Modelling
Designing Safer Cars
A computer model of a car can be used to test how safe the
design of the car is in a crash.
The virtual car can be crashed over and over again, the effects
investigated and the design easily changed until it is as safe as
possible.
This is much quicker and cheaper than building and crashing real
cars!
Weather Forecasting
A computer model of a weather system can be used to predict
storms.
The wind patterns, temperatures, etc. for the whole planet are
simulated using very powerful computers. If the computer model is
accurate (it is very difficult to make an accurate model since our
planet is rather big) then weather forecasters can use it to ‘fast-
forward’ into the future to see a prediction of what the weather
will be tomorrow, next week, next month.
(Since weather is so complex, and the models are not (yet)
accurate enough, the weather forecast can be wrong!)
Building Better Bridges
A computer model of a bridge can be used to test the design.
Bridges have to be able to survive extreme weather conditions. It
is obvious not practical to build a real bridge and then wait to see
if it falls down in a storm. Instead, a computer model of the bridge
is created and tested in virtual storms.
If the model breaks, it can be quickly and cheaply re-designed
and re-tested. If it doesn’t break, the real bridge can be built,
confident that it will survive real storms.
Bridges can also be tested to see if they can cope with heavy
traffic. The virtual bridge can be loaded with a traffic jam of virtual
trucks to check that it won’t collapse.
A similar system is used by building designers, especially for very
large or tall buildings, such as skyscrapers.
Running a Business
A computer model of a business can be used to help predict
future profits.
If the workings of a business can be modelled accurately, in
particular the financial systems, then these models can be used to
make predictions. The models are used to help answer ‘what if
…?’ type questions, e.g. “What if we decrease the workforce by
15%? Will our profits increase or decrease?”
Based on the answers that the model gives, the managers of the
business can make decisions.
Spreadsheets are often used to model the financial systems of a
business.
Training Pilots to Fly an Airplane
Trainee pilots have many hours of lessons in flight simulators before
being allowed to fly a real airplane.
Flight simulators behave almost exactly like real airplanes since
they are controlled by a computer with a very accurate and
realistic model of the airplane. The main difference is that the
simulator can’t actually crash! Pilots can make mistakes without
putting anyone’s life at risk.
Flight simulators can provide a pilot with any number of highly
realistic flying situations: storms, engine failures, low cloud hiding
the runway, etc.
The experience that pilots gain whilst using the simulator means
that when they eventually start flying real airplanes, they already
have many of the required skills.
Dangerous situations that can be tested by modelling
Natural disasters, such as hurricanes, earthquakes, and floods, can
be modeled to assess potential damage and help plan
emergency response efforts.
Traffic simulations can be used to evaluate the safety and
efficiency of different road designs, traffic signal timings, and
vehicle routing strategies.
Industrial processes, such as chemical plants and oil refineries,
can be modeled to identify potential hazards and design safer
systems.
Nuclear power plants can be modeled to predict the behavior of
the reactor and the spread of radiation in the event of an
accident.
Emergency evacuation simulations can be used to evaluate the
safety and efficiency of evacuation plans for buildings, public
transportation systems, and other public spaces.
Cyber security threats can be modeled to predict and prevent
attacks on computer networks and critical infrastructure.
Climate change can be modeled to predict the future of the
earth's environment and impact on human life.
Airline accidents can be modeled to prevent and avoid similar
accidents in the future.
There are several reasons why a computer model may be
unsuccessful:
Insufficient data: The model may not have access to enough
data, or the data may not be of high enough quality, to make
accurate predictions or decisions.
Overfitting: The model may be too complex and fit the training
data too closely, making it less able to generalize to new data.
Underfitting: The model may be too simple and not capture all the
important features of the data, making it less accurate.
Lack of domain expertise: The model may be built by individuals
without enough knowledge of the problem domain, leading to
incorrect assumptions or oversights.
Incorrect model selection: The model may not be the best choice
for the problem at hand, resulting in poor performance.
Lack of interpretability: The model may be too complex to be
easily understood and explained, making it difficult to trust its
predictions or decisions.
Lack of validation: The model may not have been properly
validated, either by using a separate dataset or by comparing
against other models, which could lead to poor performance.
Model bias: The model may be inherently biased, either in the
data or the design, which can lead to incorrect predictions or
decisions.
Lack of maintenance: The model may not be updated or
maintained over time, making it less accurate or relevant as new
data becomes available or the problem changes.
