NASA Technical Reports Server (NTRS) 20100014076: Real-Time Diagnosis of Faults Using a Bank of Kalman Filters

0.06 MB·English

by  NASA Technical Reports Server (NTRS)

Information Sciences
Conflict-Aware Scheduling Algorithm
An algorithm is being developed to automate NASA’s Deep Space Network antenna allocation.
NASA’s Jet Propulsion Laboratory, Pasadena, California
A conflict-aware scheduling algorithm scheduling algorithms that assign tracks ginning of the procedure, all requests
is being developed to help automate the that are not in conflict and mark the re- pass through a scoring system (chosen
allocation of NASA’s Deep Space Net- mainder as unscheduled. In the case from a simple mathematical equation or
work (DSN) antennas and equipment where full schedule automation is de- fuzzy logic) that determines the priority
that are used to communicate with inter- sired (based on mission/event priorities, of each request on the basis of measures
planetary scientific spacecraft. The cur- fairness, allocation rules, geometric con- of fairness of the allocation, importance
rent approach for scheduling DSN straints, and ground system capabili- of the request, the type of request, and
ground resources seeks to provide an eq- ties/constraints), a conflict-free schedule the allocation length. Starting with the
uitable distribution of tracking services can easily be created from the conflict- highest priority request, all technical
among the multiple scientific missions aware schedule by removing lower prior- and geometric constraints are combined
and is very labor intensive. Due to the ity items that are in conflict. to determine the available “timeline/an-
large (and increasing) number of mis- Unlike most existing scheduling en- tenna groups” for scheduling. A scoring
sion requests for DSN services, combined gines that require fixed length schedule system that considers items already in
with technical and geometric constraints, items in the request, the conflict-aware the schedule and the request character-
the DSN is highly oversubscribed. To schedule provides a dynamic scheduling istics then identifies the best
help automate the process, and reduce engine to determine allocation length timeline/antenna group and start times
the DSN and spaceflight project labor ef- during the scheduling process. This is for each request. This then continues
fort required for initiating, maintaining, made necessary by the variety of mission- for each successive priority request (pri-
and negotiating schedules, a new sched- tracking request types faced by the DSN. ority is recomputed dynamically) until
uling algorithm is being developed. In addition to fixed track requests, mis- all requests are scheduled.
The scheduling algorithm generates a sions may also need continuous coverage The conflict-aware algorithm is not
“conflict-aware” schedule, where all re- or may need to segment a track related to limited to DSN application. It can also
quests are scheduled based on a dynamic multiple ground assets to support a given be applicable to solution of scheduling
priority scheme. The conflict-aware request for service. In these cases, the problems in areas such as manufactur-
scheduling algorithm allocates all re- schedule allocation length (time) will de- ing and traffic control.
quests for DSN tracking services while pend on the availability of each resource. This work was done by Yeou-Fang Wang
identifying and maintaining the conflicts The conflict-aware scheduling algo- and Chester Borden for NASA’s Jet Propul-
to facilitate collaboration and negotia- rithm combines scheduling heuristics, sion Laboratory. Further information is
tion between spaceflight missions. These optimization, a search algorithm, and contained in a TSP (see page 1).
contrast with traditional “conflict-free” computational intelligence. At the be- NPO-41320
Real-Time Diagnosis of Faults Using a Bank of Kalman Filters
Gradual changes associated with aging are taken into account in the diagnostic process.
John H. Glenn Research Center, Cleveland, Ohio
A new robust method of automated detecting component faults (abrupt count for gradual changes in perform-
real-time diagnosis of faults in an aircraft degradation in engine component per- ance associated with aging of an other-
engine or a similar complex system in- formance). By affording a capability for wise healthy engine. By incorporating
volves the use of a bank of Kalman filters. real-time identification of minor faults be- information on gradual, aging-related
In order to be highly reliable, a diagnostic fore they grow into major ones, the changes, the new method makes it possi-
system must be designed to account for method promises to enhance safety and ble to retain at least some of the sensitiv-
the numerous failure conditions that an reduce operating costs. ity and accuracy needed to detect incipi-
aircraft engine may encounter in opera- The robustness of this method is fur- ent faults while preventing false alarms
tion. The method achieves this objective ther enhanced by incorporating infor- that could result from erroneous inter-
though the utilization of multiple Kalman mation regarding the aging condition of pretation of symptoms of aging as symp-
filters, each of which is uniquely designed an engine. In general, real-time fault di- toms of failures.
based on a specific failure hypothesis. A agnostic methods use the nominal per- The figure schematically depicts an
fault-detection-and-isolation (FDI) system, formance of a “healthy” new engine as a FDI system according to the new
developed based on this method, is able to reference condition in the diagnostic method. The FDI system is integrated
isolate faults in sensors and actuators while process. Such an approach does not ac- with an engine, from which it accepts two
NASA Tech Briefs, January 2006 29
ing the residual output of each filter, the
specific fault that has occurred can be
detected and isolated on the basis of the
Output: decision rules.
+
Sensor Bank of FDI Indication of a A set of parameters that indicate the per-
Readings Kalman Decision Fault in a Sensor, or
– Filters Rules Actuator, or Other formance of the engine components is es-
Engine Component timated by the “correct” Kalman filter for
use in detecting component faults. To re-
– duce the loss of diagnostic accuracy and
Data Bank Updates of sensitivity in the face of aging, the FDI sys-
of + Kalman
Trim Values Filters or tem accepts information from a steady-
Trim Values state-condition-monitoring system. This in-
formation is used to update the Kalman
Actuator Information From filters and a data bank of trim values repre-
Commands Steady-State-Condition-Monitoring
sentative of the current aging condition.
System
This work was done by Takahisa
Kobayashi of QSS Group, Inc. and Donald L.
Simon of Army Research Laboratory for
This FDI Systemaccepts information on gradual aging of an engine to retain its effectiveness in Glenn Research Center. Further informa-
fault diagnosis. tion is contained in a TSP (see page 1).
Inquiries concerning rights for the commer-
sets of input signals: sensor readings and designed to detect a specific sensor or ac- cial use of this invention should be addressed
actuator commands. Two main parts of tuator fault. When a sensor or actuator to NASA Glenn Research Center, Commercial
the FDI system are a bank of Kalman fil- fault occurs, large estimation errors are Technology Office, Attn: Steve Fedor, Mail
ters and a subsystem that implements generated by all filters except the one Stop 4-8, 21000 Brookpark Road, Cleveland,
FDI decision rules. Each Kalman filter is using the correct hypothesis. By monitor- Ohio 44135. Refer to LEW-17457.
LEW-17457
ABPI
12-15-03 es
30 NASA Tech Briefs, January 2006