3: The Implementation of a Structural Style Identification Expert System
Structural style analysis while providing the basis for determining the tectonic evolution and framework of a basin also helps seismic interpreters in selecting appropriate structural models to guide interpretation of profiles and preparation of maps. Research results on structural style identification have been documented in many conference and journal papers. If a thorough knowledge of the identification criteria and a systematic application of this knowledge could be captured using the expert systems approach, this expertise could be widely distributed.
An expert system on structural style identification was built on an HP-320 AI workstation. The knowledge base contains map and profile criteria for identifying a style, differentiating a style from possible pitfalls, and refuting pitfalls. The identification procedure presents the user with general profile characteristics to nominate a potential style. This nomination is checked with detailed identification criteria. If the checking is positive, then pitfalls are tested and refuted. A confirmation is reached if the support of the nomination is strong and if all pitfalls either are tested negative, or are refuted. Otherwise, an alternate style is nominated, and the same procedure is repeated.
A commercial expert system shell is used to organize and test the acquired knowledge base. The knowledge base is separated into three parts: procedure, criteria, and certainty combination rules. They are treated as strategic, factual, and judgmental knowledge, respectively. This segregation of knowledge makes the development of the knowledge base very adaptive to the frequent editorial changes made by the domain expert.
In the final system, the expert system shell serves as an inference engine. The system also contains two other major components: a user interface and a data base manager. The user interface is completely mouse driven and is built upon the X-window system, which also provides the communication between the user, the inference engine, and the data base. All the style identification criteria and their corresponding sketches are stored in the data base. At each stage of consultation, all the criteria pertinent to a structural style are listed in a panel. By clicking the mouse on the selection box of each of these criteria, the data base manager retrieves corresponding sketches and displays them in a different window. In the meantime, the inference engine deduces an intermediate decision and moves on to the next stage of consultation. At the end of a consultation, the user can save the consultation in the case history file in the data base.