The SHINE Knowledge Base Interchange Language software has been designed to more efficiently send new knowledge bases to spacecraft that have been embedded with the Spacecraft Health Inference Engine (SHINE) tool. The intention of the behavioral model is to capture most of the information generally associated with a spacecraft functional model, while specifically addressing the needs of execution within SHINE and Livingstone. As such, it has some constructs that are based on one or the other.
As NASA/JPL autonomous science missions go deeper and deeper into space, the collection of unexpected data becomes a problem. Data structures can easily be implemented in advance that can collect any kind of data; however, when it comes to processing the data into information and taking advantage of serendipitous science discovery, designing a fixed and efficient data structure becomes increasingly complex. This software defines and implements a new kind of data structure that can be used for representing information that is derived from serendipitous data discovery. It allows the run-time definition of arbitrarily complex structures that can adapt at run-time as the raw science data is transformed into information.
This solves the problem decision trees can be prone to, namely how expensive they can be to execute because of the need to evaluate each non-leaf node and, based upon its truth, to either progress deeper into the structure or to examine an alternative. This requires many machine cycles, which can negatively affect time-critical decisions.
This software runs on a variety of different platforms, including SUN, HP, Intel, Apple Macs, Flight Processors, etc. It can be distributed in either source code or binary code and requires a LISP compiler to run with a number, such compilers being either commercially available or found as shareware. The software has no specific memory requirements and depends on the applications that are running in it. It is implemented as a library package and folds into whatever environment is calling it.
Currently, this software is a component of the Common Automation Engine (CAE) that was developed for Deep Space Network (DSN). It has been in active use for over three years and has been installed in a shadow mode running at Goldstone and DSN monitoring operations at JPL.
This work was done by Mark James, Ryan Mackey, and Raffi Tikidjian of Caltech for NASA's Jet Propulsion Laboratory.
This software is available for commercial licensing. Please contact Karina Edmonds of the California Institute of Technology at (626) 395-2322. Refer to NPO-44546.
This Brief includes a Technical Support Package (TSP).
XML-Based SHINE Knowledge Base Interchange Language
(reference NPO-44546) is currently available for download from the TSP library.
Don't have an account?
Overview
The document outlines the development of an XML-based knowledge base interchange language, specifically designed for the Spacecraft Health Inference Engine (SHINE) and related applications. This initiative, identified as NTR-44546, aims to standardize the representation and exchange of decision trees, which are crucial for encoding knowledge-based problems in software systems, particularly in aerospace contexts.
Decision trees are hierarchical structures that utilize true/false responses to navigate through a set of queries, ultimately leading to conclusions. They are prevalent in various software systems, including those used by NASA for control and diagnostics. However, executing decision trees can be computationally expensive, especially in time-critical scenarios, necessitating an efficient method for their implementation.
To address these challenges, the document introduces SHINEXML, a sublanguage defined in XML that allows for the encoding of complex decision tree-based behavioral models. The structure of SHINEXML includes constructs for defining decision trees and decision nodes, enabling the specification of tests and actions based on the outcomes of those tests. The language facilitates the creation of a decision tree that can be transformed into a set of forward chaining inference rules, which are essential for the operational efficiency of SHINE.
The document also describes the process of compiling SHINEXML into a target language, SHINE, through a transformational compiler. This compiler applies a series of transformational rules to map SHINEXML statements into a format compatible with SHINE, enhancing the decision-making capabilities of the system.
In summary, the document presents a comprehensive approach to encoding decision trees using a standardized XML language, addressing the need for efficient knowledge base interchange in aerospace applications. By leveraging SHINEXML, the initiative aims to improve the execution of decision trees, making them more suitable for real-time applications in spacecraft health monitoring and diagnostics. This innovative method not only enhances the operational capabilities of SHINE but also sets a precedent for future developments in knowledge representation and exchange across various systems.
