Bernhard Rinner and Benjamin Kuipers. Monitoring Piecewise Continuous Behaviors by Refining SemiQuantitative Trackers. In Proceedings of the Sixteenth International Joint Conference on Artificial Intelligence (IJCAI-99), pages 1080--1086, Stockholm, Sweden, August 1999. Morgan Kaufmann.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....places, in VLSI simulation [28] To date, the bounds for such functions have been derived in an ad hoc manner. The work described in this paper provides a systematic method for finding these functional bounds. It is particularly appropriate for semiquantitative monitoring and diagnosis systems[10, 25, 26, 27]) for chemical plants or devices such as turbine engines because process data is readily available in such applications. 2 Computing the Estimate Computing the estimate of g requires that we make some assumptions about the nature of deterministic and stochastic portions of the model. We assume ....

Bernhard Rinner and Benjamin Kuipers. Monitoring Piecewise Continuous Behaviors by Refining Trackers and their Models. Proceedings of the Sixteenth International Joint Conference on Artificial Intelligence, Stockholm, Sweden 1999.

....be estimated. Further, the candidate qualitative diagnoses were generated from initial observations of aberrant behavior, and their consistency can be further evaluated by monitoring the qualitative transients associated with each candidate. The refinement process is performed by a set of trackers [21], one for each candidate diagnosis and associated model. Each tracker comprises both a qualitative transient analysis component and a quantitative model estimation, component. The two components operate in parallel as described below. Qualitative Transient Analysis The qualitative transient ....

B. Rinner and B. Kuipers. Monitoring piecewise continuous behavior by refining trackers and models. In Hybrid Systems and AI: Modeling, Analysis and Control of Discrete + Continuous Systems, AAAI Technical Report SS-99-05, pp. 164--169, 1999.

No context found.

Bernhard Rinner and Benjamin Kuipers. Monitoring Piecewise Continuous Behaviors by Refining SemiQuantitative Trackers. In Proceedings of the Sixteenth International Joint Conference on Artificial Intelligence (IJCAI-99), pages 1080--1086, Stockholm, Sweden, August 1999. Morgan Kaufmann.

....are grounded on the refutation of subspace models that are known to be inconsistent with the measurements. Semi quantitative system identification performs refinement at the qualitative and interval level. Semi quantitative system identification has also been applied to model based monitoring [11]. Bonarini and Bontempi [4] have developed a quite similar approach to our consistency check. However, they have focused on uncertainty initial state values, which are given as intervals. Also related to our work is Armengol et al. 1, 2] The simulation is based on modal interval arithmetics, ....

Bernhard Rinner and Benjamin Kuipers, `Monitoring Piecewise Continuous Behaviors by Refining Semi-Quantitative Trackers', in Proceedings of the Sixteenth International Joint Conference on Artificial Intelligence (IJCAI-99), pp. 1080--1086, Stockholm, Sweden, (August 1999). Morgan Kaufmann.

....detected by considering the range of a laser reading as a function of the angular sector s length associated with the scan. In order to identify monotonic (i.e. increasing, decreasing, or constant) regions in the data we adapted the qual itative filtering algorithm proposed in [Kay et al. 1999, Rinner and Kuipers, 1999] This algorithm uses a window of k window size consecutive readings to determined the different trends in the data. Whether a trend is declared to be increasing, decreasing or constant, is determined by comparing the slope of points in a window against a threshold k std dev. For each cluster in ....

B. Rinner and B. Kuipers. Monitoring piecewise continuous behaviors by refining trackers and theirs models. In Sixteenth International Joint Conference on Artificial Intelligence, pages 1191-1198, August 1999.

....detected by considering the range of a laser reading as a function of the angular sector s length associated with the scan. In order to identify monotonic (i.e. increasing, decreasing, or constant) regions in the data we adapted the qualitative filtering algorithm proposed in [Kay et al. 1999, Rinner and Kuipers, 1999] This algorithm uses a window of k window size consecutive readings to determined the different trends in the data. Whether a trend is declared to be increasing, decreasing or constant, is determined by comparing the slope of points in a window against a threshold k std dev. For each cluster in ....

B. Rinner and B. Kuipers. Monitoring piecewise continuous behaviors by refining trackers and theirs models. In Sixteenth International Joint Conference on Artificial Intelligence, pages 1191--1198, August 1999.

.... Qualitative Simulation DRAFT: February 26, 2001 18 (e.g. the Three Mile Island nuclear plant failure) often occurs due to operator xation on a single hypothesized model of the system that is only refuted after it is too late to x a developing problem [26] The MIMIC approach to monitoring [10, 28] tracks multiple hypotheses in parallel, each expressed as an SQDE model. Any desired features in the observation stream can be used to trigger fault hypotheses, launching additional trackers to run in parallel, even before the nominal model is refuted. Multiple active trackers and their ....

Bernhard Rinner and Benjamin Kuipers. Monitoring piecewise continuous behaviors by rening semiquantitative trackers. In Proceedings of the Sixteenth International Joint Conference on Articial Intelligence (IJCAI-99), Stockholm, Sweden, 1999.

....t u ; t s t u ] Figure 13(b) Note that for a valid mapping the intersection between SQ trend and SQ behavior must be non empty at any time o set within the time uncertainty. This precondition can be exploited to narrow the time uncertainty before the actual trend behavior mapping takes place [33]. Time uncertainty results in broader numerical bounds in the trajectory space and, therefore, in less e ective re nements. However, the mapping process between SQ trend and SQ behavior remains conservative and no modi cations are needed for the model re nement step. 2 The overlap may be ....

Bernhard Rinner and Benjamin Kuipers. Monitoring Piecewise Continuous Behaviors by Rening SemiQuantitative Trackers. In Proceedings of the Sixteenth International Joint Conference on Articial Intelligence (IJCAI-99), pages 1080-1086, Stockholm, Sweden, August 1999. Morgan Kaufmann.

....t u ; t s t u ] Figure 15(b) Note that for a valid mapping the intersection between SQ trend and SQ behavior must be non empty at any time offset within the time uncertainty. This precondition can be exploited to narrow the time uncertainty before the actual trend behavior mapping takes place [34]. Time uncertainty results in broader numerical bounds in the trajectory space and, therefore, in less effective refinements. However, the mapping process between SQ trend and SQ behavior remains conservative and no modifications are needed for the model refinement step. 4 Experimental Results ....

Bernhard Rinner and Benjamin Kuipers. Monitoring Piecewise Continuous Behaviors by Refining Trackers and their Models. In Proceedings of the Sixteenth International Joint Conference on Artificial Intelligence, Stockholm, Sweden, August 1999. Morgan Kaufmann.

No context found.

B. Rinner and B. Kuipers. Monitoring piecewise continuous behavior by refining trackers and models. In Hybrid Systems and AI: Modeling, Analysis and Control of Discrete + Continuous Systems, AAAI Technical Report SS-99-05, pp. 164--169, 1999.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC