Diagnostics and a qualitative model

Abstract

First generation expert systems were using shallow knowledge based on heuristic information to solve a diagnostic problem. This approach has many disadvantages, which can be avoided by using deep knowledge. Diagnostic reasoning based on deep knowledge is called model-based diagnostics. Recently, the use of qualitative modeling in relation to deep knowledge in expert systems has become increasingly important. The main purpose of our contribution is to present the model-based diagnostic approach at a formal level. The originality of the presented formalization is the concept of the diagnostic space, the characterization of the minimal diagnoses, and the measurement. The formalization serves as the theoretical background to prove our view to the design of qualitative system models and to establish the diagnostic architecture called DISY. The qualitative system model in our diagnostic approach needs not to be specially adopted for use in the diagnostic domain. The only requirement is that it must simulate the system behavior expressed by normal or abnormal functioning of its components. Proposed DISY architecture is not complex and simply takes into an account the previous diagnostic result to obtain a new one from the additional observation-measurement (medical tests or examinations) of the system.

Introduction

A diagnostic problem is a determination of those components of a system, which, if assumed to be functioning abnormally, will explain the discrepancy between the observed and the correct system behavior [1].

Diagnostic reasoning in the first generation expert systems was using shallow knowledge based on heuristic information to solve a diagnostic problem. However, the first generation expert systems have significant difficulties expressing the knowledge that is necessary to understand the interactions between the irregularities. The fundamental problem is that they do not understand the processes that underlie the observable findings. Therefore, they do not understand how the mechanisms in their irregular states function to produce the observed behavior [2].

To avoid these disadvantages, we need an approach that can explain how the normal mechanism responds to different changes and how a faulty mechanism might produce the observed manifestations. These demands can be met by using deep knowledge. The only available information in this case is the system description (its design or structure), together with the observation of the system behavior. This type of diagnostic reasoning is called model-based diagnostics.

The result of diagnostic reasoning is a set of declarations about components’ behavior called diagnosis, which could explain unexpected system behavior. Generally, there may be several competing diagnoses, which explain the same faulty system. However, only a unique diagnostic result ensures a proper intervention on the system to restore its normal behavior, what is the final goal of diagnostics. To discriminate among all competing diagnoses, an additional observation also called measurement (medical test or examination) is needed. For efficiency reasons, it is necessary to consider the previous diagnoses for computing new diagnoses for a given new measurement.

A theory of diagnoses presented in Reiter's framework [1] is adopted as the fundamental theoretical contribution to the model-based diagnosis. Hou later proposed a general theory of measurement in model-based diagnoses and provided a formal justification for his theory [3]. We propose a diagnostic architecture DISY for computing new diagnoses with a given new measurement, based on the previous diagnoses, adjusted for simulation process using the system model. For a formal argumentation of the presented diagnostic architecture and the role of the system model, we developed an original theoretical framework of a diagnostics.