Session
Technical Session IIB: Advanced Operations Concept Systems
Abstract
A Perl Inference Engine (PIE), A Plan Executor (APE), and a Centroid Classifier (CCL) are described within the infrastructure provided by the Knowledge Server Toolkit (KST) environment. All these tools are implemented as Perl modules so that user's applications can easily use their capabilities. This technology was developed at NASA Goddard Space Flight Center under a task that was looking into languages and tools that are useful for automating satellite ground control systems. However, because these tools are modularized and knowledge driven, this technology should be generally applicable to many fields. After a background description, the Knowledge Representation (KR) and search strategy used by PIE are described. PIE applications are described next and include a simple data checker filter and a PIE Knowledge Server (PIEKS). Then, some simple monitors that display raw and inferred data are also described. These include real-time displays of alpha-numeric data and plots of raw and calculated data. Next, the environment of APE (APE is a client of PIEKS) and its KR are described. A description of a CCL server follows. CCL allows the user to define fuzzy classifier applications and therefore the code should prove useful as a template for writing neural net classifier servers. Finally, the application of KST to automating the Earth Observing System Data and Operations System (EDOS) is briefly described.
Knowledge Server Toolkit: Perl-based Automation Tools
A Perl Inference Engine (PIE), A Plan Executor (APE), and a Centroid Classifier (CCL) are described within the infrastructure provided by the Knowledge Server Toolkit (KST) environment. All these tools are implemented as Perl modules so that user's applications can easily use their capabilities. This technology was developed at NASA Goddard Space Flight Center under a task that was looking into languages and tools that are useful for automating satellite ground control systems. However, because these tools are modularized and knowledge driven, this technology should be generally applicable to many fields. After a background description, the Knowledge Representation (KR) and search strategy used by PIE are described. PIE applications are described next and include a simple data checker filter and a PIE Knowledge Server (PIEKS). Then, some simple monitors that display raw and inferred data are also described. These include real-time displays of alpha-numeric data and plots of raw and calculated data. Next, the environment of APE (APE is a client of PIEKS) and its KR are described. A description of a CCL server follows. CCL allows the user to define fuzzy classifier applications and therefore the code should prove useful as a template for writing neural net classifier servers. Finally, the application of KST to automating the Earth Observing System Data and Operations System (EDOS) is briefly described.
