Task-dependent qualitative domain abstraction

Abstract

Automated problem-solving for engineered devices is based on models that capture the essential aspects of the behavior. In this paper, we deal with the problem of automatically abstracting behavior models such that their level of granularity is as coarse as possible, but still sufficiently detailed to carry out a given behavioral prediction or diagnostic task. A task is described by a behavior model, as composed from a library, a specified granularity of the possible observations, and a specified granularity of the desired results. The goal of task-dependent qualitative domain abstraction is to determine maximal partitions for the variables' domains (termed qualitative values) that retain all the necessary distinctions. We present a formalization of this problem within a relational (constraintbased) framework, and devise solutions to automatically determine qualitative values for a device model. The results enhance the ability to use a behavior model of a device as a common basis to support different tasks along its life cycle.

Parametric abstraction of behavioral modes for model-based diagnosis

“…Some recent works in MBD (see, e.g., [5,19,20,28]) have pointed out that the main driving criterion for performing an abstraction is the level of observability of the system, and in particular the impossibility of discriminating among behavioral modes of one or more components: a low level of observability usually leads to a large number of solutions for a diagnostic problem.…”

Model-Based Diagnosis Meets Combinatorial Testing For Generating an Abductive Diagnosis Model

“…Of course we have to take care when designing the alphabet as of Definition 4 for these variables. However, available work from Sachenbacher and Struss can help in finding an appropriate abstraction in an automated fashion [21]. Please note that in most cases, we could certainly define these variables' alphabets as sets of elements rather than sets of sequences.…”

“…That is, such variables state, for example, a component's finite parameters or a comparison to a constant such as to digitize the value (which is also in line with the input parameter modeling task mentioned for combinatorial testing). For a discussion of how to identify an appropriate task dependent qualitative abstraction and a corresponding automated approach, we refer the interested reader to [21].…”

“…When it comes to continuous signals and temporal behavior, obviously the input space for these variables grows significantly. The before-mentioned input modeling task is one where we should definitely take care thus, possibly drawing on work on automated abstraction like [21]. For defining our algorithms, we took an approach where we could define several individual patterns for each variable, to be chosen from in the generation process.…”

The qSafe Project – Developing a Tool for Current Practice in Functional Safety Analysis

“…This comprises, in particular:  Generation of qualitative models by automated discretizing of implemented numerical models  Automated generation of (qualitative) deviation models from equation-based, numerical models  Generation of models for the derivation of effects over time (e.g. system response after an error)  Task-dependent behavior abstraction: coarsening of a system model without losing the force of expression regarding the task given by system, requirements and scenarios [29].…”