This research work has started in 1988 and continues. It is done mainly in collaboration with Associate Professor Ashok K. Goel.
This work has focussed on two basic issues in reusing the problem
solving knowledge of an intelligent system for a new problem.
The adaptive modeling methodology is used for design problem
solving in the domain of engineering devices. In order to capture the
knowledge of designing a device, several Functional Models are
developed. These models map the physical attributes of devices onto the
goals of the designer. This work is conducted through several projects
which are listed below.
- Adaptive Modeling:Development of
new representations that can capture problem solving knowledge such
that it can be reused for solving new problems.
- Adaptation Strategies: Development of new strategies
that search through a space of adaptations of problem solving knowledge.
This results in transferring the knowledge of solving an old problem to
new problem situations.
This work has been conducted at the Laboratory for Artificial Intelligence Research, The Ohio State University, in collaboration with Associate Professor Ashok Goel. In this work, the functional model is a Structure, Behavior and Function (SBF)
model. The adaptation strategies generate and search through an
adaptation space of these SBF models. The resulting SBF model is the
preliminary design that addresses the new design problem.
This work simulates the interaction between the device and its
environment for the device failure to deliver its functionality in the
environment. This is done by using the SBF models of the device and the
environment. This simulation is analysed and faults are diagnosed within
the SBF models. These faults are repaired by a compositional modeling
Many devices are perceived as delivering their functionalities by
interacting with their environments. This work develops a new
representation called Environmental SBF (ESBF) model. It also presents
adaptation strategies that adapt the ESBF models of some known devices
to design new ESBF models for new device and environmental interactions.
This work focuses upon how to reuse the process of designing a device for designing new devices. We developed a new representation called Methods, Strategies and Tasks (MST)
model which represents the process of designing from first principles.
The states of this representation are the intermediate devices that are
modeled using SBF model. The adaptation strategies adapt both the MST
and SBF models to arrive at new device models.
Designers are often required to do diagnosis on the devices they have
designed. This requires that the models they used for designing need to
be used for diagnosing and repairing these devices. We developed a
strategy that reformulates an SBF model for a failure such that
diagnosis and repair tasks can be performed on it to eliminate the
The strategies present within the MST model may not be adequate for a
design problem. Direct adaptation may not result in a required strategy.
In this work, we use a meta-strategy that reformulates the MST model of
a device such that it can embed a required strategy.
In order to capture the knowledge of designing a device, so that it can be reused to design new devices, several Functional Models
are developed. These models map the physical attributes of the devices
onto the goals of the designer. These functional models include Structure, Behavior and Function (SBF) models, Environmental Structure, Behavior and Functional(ESBF) models and Methods, Strategies and Task (MST) models.
Diagnosing multiple-diseases in human body is a very difficult task to
model. This work addresses this problem by simulating the disease
progress within the body by using a set of prototypical qualitative
disease processes. Given symptoms, these processes are selected and a
disease progress is simulated. The simulation that explains the largest
number of symptoms present in the patient is the diagnosis.
Updated on February 20, 2002