Measuring and testing – Simulating operating condition – Marine
Reexamination Certificate
1998-12-28
2003-04-01
McCall, Eric S. (Department: 2855)
Measuring and testing
Simulating operating condition
Marine
Reexamination Certificate
active
06539783
ABSTRACT:
BACKGROUND OF THE INVENTION
This invention relates generally to aircraft engines and more particularly, to a multipoint engine model and tracking filter for estimating engine health.
Engine components typically are scheduled for maintenance based on a preselected number of operation hours or cycles. The preselected number typically is conservatively selected based on numerous factors including past component experience. If a component fails, a predetermined diagnosis routine is followed to identify and replace the failed component.
To estimate engine health, it is customary to model values such as engine efficiency and engine corrected flow. These modeled characteristics, however, are based on a nominal or average new engine. Due to manufacturing tolerances, faults, damage, or deterioration with time, actual engine characteristics typically are different from the assumed nominal characteristics.
Rather than simply performing maintenance based on a preselected number of operation hours or cycles, or modeling engine performance based on a nominal new engine, it would be desirable to track deterioration of the actual engine components over time. Algorithms for tracking engine health are sometimes referred to herein as tracking filters, sand such tracking filters provide “health estimates” of engine components.
Health estimates can be used to facilitate “on condition” maintenance rather than maintenance based on a number of hours or cycles. Performing maintenance based on actual engine health facilitates eliminating unnecessary maintenance. In addition such tracking facilitates fault diagnosis in that the status of each component would be known and, based on the actual known state of each component, the failed component can be readily identified.
The health estimates also can be used to improve computation of sensed parameters (e.g., rotor speeds, temperatures, and pressures). Specifically, the condition of the components is indicative of such parameters. Health estimates therefore provide sensor redundancy and even the potential to eliminate sensors by using “virtual” sensors. Further, health estimates can be used to improve computation of parameters that are not sensed, such as thrust, turbine inlet temperature, stall margins, and air flows. These parameters can be used by the control logic to, for example, reduce required margins and to improve engine performance.
Known tracking filters do not consider information from more than one operating point simultaneously, and the number of parameters estimated is equal to the number of sensors. Since the number of sensors is usually much smaller than the number of parameters to be estimated, such filters combine the effect of several parameters into a few parameters, which inhibits individually tracking each parameter.
Examples of known filters include steady-state tracking filters, dynamic tracking filters using, for example, Kalman tracking filters, and least-squares estimator filters. Nonlinear estimation filters including neural networks or fuzzy rule-based systems also can be used. Although known filters perform reasonably well when estimating secondary parameterst required for model based control, such filters are not well suited for model based diagnostics.
The least-squares tracking filter, for example, solves a system in which the number of observations (sensors) is smaller than the number of unknowns (parameters). However, the least-squares estimate, in general, differs from the actual component characteristics. Specifically, there are fewer sensors than parameters, and the system is not “observable” using sensor data from a single operating condition. To improve engine health filter performance, it would be desirable to increase observability, i.e., increase sensor data.
BRIEF SUMMARY OF THE INVENTION
These and other objects may be attained by a tracking filter system that matches all model parameters to corresponding engine parameters and therefore, the model matches the actual engine. The tracking filter system provides that the model tracks the engine at steady-state as well as dynamically, and does not adversely affect control stability and robustness. More specifically, the present filter obtains information from two or more engine operating conditions to estimate health parameters P, performance parameters Y
p
, and sensor estimates Y
s
. The operating condition is defined by a combination of flight conditions and engine power level. The information from each operating point is considered to be linearly independent. The nonlinearily inherent in the engine is therefore leveraged rather than viewed as a problem handled by the control system or estimation scheme.
By collecting sensor data from a multiplicity of linearly independent operating conditions before using the data for parameter estimation, observability is increased because sensor data from n-operating conditions simultaneously is similar to having n times the number of sensors. Therefore, the number of observations (sensors) is greater than the number of unknowns (parameters), which results in good estimates of the parameters, and tolerance to sensor biases.
In one specific implementation, a tracking filter compensation unit is coupled to receive corrected sensor measurements from the engine. The sensor measurements are combined with previously generated model estimates, and the combined sensor measurements and model estimates are provided to the compensation unit. The compensation unit generates filter updates which are supplied to an engine model unit. The engine model unit also receives corrected environmental sensor inputs and control signals from an engine control unit. The engine model unit generates updated model estimates which are then subsequently supplied to the tracking filter compensation unit as described above.
The engine model unit obtains information from two or more operating conditions to estimate health parameters P, performance parameters Y
p
, and sensor estimates Y
s
. Using such multipoint estimation of engine health enables estimation of all engine health parameters rather than only a few engine health parameters for use in engine component diagnostics, trending, and fault detection and isolation. In addition, such estimation is believed to provide improved estimates of sensed parameters for use in sensor redundancy management and sensor elimination, and also is believed to provided improved estimates of performance parameters for use in model based control to reduce required temperature, stall, and other margins.
REFERENCES:
patent: 4215412 (1980-07-01), Bernier et al.
patent: 5080496 (1992-01-01), Keim et al.
patent: 5489829 (1996-02-01), Umida
patent: 5689066 (1997-11-01), Stevenson
patent: 5726891 (1998-03-01), Sisson et al.
patent: 858017 (1998-08-01), None
patent: 0858017 (1998-08-01), None
Armstrong Teasdale LLP
General Electric Co.
Herkamp Nathan D.
McCall Eric S.
Reeser III Robert B.
LandOfFree
Methods and apparatus for estimating engine health does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Methods and apparatus for estimating engine health, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Methods and apparatus for estimating engine health will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3064297