Project Summary
CESADS/INTELMOD Project is an ESA/ESOC program with the objective of demonstrate the advantages and limitations of using “Knowledge Engineering” to support multi-ground stations, namely in: System-level health status monitoring, Failure recognition and diagnostic, Failure recovery & System reconfiguration, and Operational performance monitoring & diagnostic. The tool developed in the framework of this project will be installed and made available to the operators for use and evaluation.The project is split into two parts though closely related: the Knowledge Engineering one that intends to first understand where exactly a knowledge-based system can be useful in ESOC context (i.e. problems and opportunities), and then to gather all the require experts’ knowledge in that sense; finally, the second part will consist in designing and implementing the referred knowledge-based system. Knowledge Engineering A key point in the success of this endeavour is the choice of following a knowledge engineering methodology, in this case the CommonKADS one. CommonKADS is a result of the merge of both scientific and practical expertise in the field of knowledge engineering, which focus several aspects, from the knowledge elicitation sessions to knowledge management and knowledge systems design. Thus, in CESADS scope, CommonKADS has been used to organize the knowledge elicitation sessions, to organize the acquired knowledge, and to model the knowledge (CommonKADS provides a set of templates).
Concerning the results, it has been possible to undercover the most important problems in ESOC G/S operations, such as space link maintenance and the integration, performed by the operators in stress situations, of different heterogeneous information sources (e.g. S/C telemetry, G/S and network status), so allowing us to drive the remaining investigations. In the scope of those problems, interviews and literature has been consulted to understand the inferences performed by the operators during their activities.A direct outcome of applying knowledge engineering as analysis tool has been the formalization of knowledge present in ESOC operations; thus, knowledge management is also provided as a service of the project, which pushes the outcomes of this project far beyond its conclusion. Knowledge-Based System The Knowledge-Based System to be developed in CESADS scope is based on the Multi-Agent Knowledge-Based Architecture that UNINOVA and GTD have been cooperatively developing. Such architecture will: - Provide knowledge acquisition tools to enable experts to add, manage, and refine ontology and inference knowledge present in the system.
- Integration of heterogeneous and distributed information
- Storage, Analysis, and Reporting features over historical data (DW + OLAP)
- Data Mining environment
- Real-Time monitoring and diagnosis of the interfaced systems, integrating
- Symbolic Reasoning (e.g. rule-based, CBR) and
- Non-Symbolic Methods (e.g. predictions, classification)
- Domain oriented, and adaptive HMIs
- The agent concept beneath this architecture provides flexibility, adaptation, robustness, and maintenance easiness of the system, since there is no single point of failure, the system degrades progressively, and allows several “artificial” experts reasoning about the same issue.
Figure 2 depicts the current scenario at ESOC (left), whereas all systems are heterogeneous making difficult to integrate the data. Moreover, operators at ESOC have to deal with all the systems at the same time, which means interfacing different systems and performing correlations by experience (tacit knowledge). CESADS will interface some of those systems and integrate their information by means of developing a common ontology. Therefore, in the future (right) the experts and users will be able to interface all systems from a single gateway, with a common (domain oriented) vocabulary.
A great challenge in this project is that a S/C pass is far from being a sequence of a set of homogeneous tasks; it is clear that during a pass, operators have to deal with different time constraints, which means that CESADS has also to deal with those situations as well. Furthermore, the interactions between several actors at ESOC to perform a single task shows how distributed and complex the organization is, which is in fact the perfect scenario for a multi-agent system.
Research Areas
- Knowledge Engineering
- Knowledge-Based Systems (Expert Systems, Ontology)
- Predictive Models
Partners
Publications Ricardo Raminhos, António Falcão, Luís Correia, “CESADS – Centralised ESTRACK Status and Diagnostic System”, World Automation Congress (WAC) – Special Session on “Soft Computing in Space Applications”, Budapest, Hungary, 2006
Links
|
