Means and method for system performance tracking
First Claim
1. A method of monitoring changes in an engine, said method specifically monitoring n performance parameters of said engine by means of n performance parameters changes in the form of a multidimensional vector dP, said engine having a plurality of control inputs and resulting in m engine outputs, where n and m are positive integers with n greater than m, said method comprising the steps of:
- (i) providing a computer model of the engine, said model having a plurality of control inputs and having m model outputs;
(ii) initializing said multidimensional vector dP to zero;
(iii) storing at least one non-square matrix C of sensitivity coefficients which relates changes in performance parameters of the engine dP to a vector of differences between engine sensor outputs and computer model outputs dx, where dx=CdP;
(iv) operating said engine;
(v) computing said m model outputs based upon said plurality of control inputs and said vector of performance chances dP;
(vi) comparing said computed m model outputs to m sensor engine outputs and calculating a vector of output sensor changes dx;
(vii) applying singular value decomposition to compute a revised dPe, where C=USS ;
0VT and VR (SS)-1 UT is the pseudoinverse of C.
2 Assignments
0 Petitions
Accused Products
Abstract
A method of tracking a system (1) having sensors (3) (e.g. gas turbine engines) by providing a computer model (4) which, from changes in sensor readings, can track (follow) changes in performance parameters of the system. These may be e.g. efficiencies of various of components such as compressors and turbines. In the common embodiment, the method used (Singular Value Decomposition) is a way of computing the most likely solution when there are more performance parameters than the number of sensors used and thus a degree of redundancy. The method allows system diagnosis and most importantly a tracked model allows the selection of a tailored control regime (9) which results in optimum efficiency for the individual system (engine).
-
Citations
9 Claims
-
1. A method of monitoring changes in an engine, said method specifically monitoring n performance parameters of said engine by means of n performance parameters changes in the form of a multidimensional vector dP, said engine having a plurality of control inputs and resulting in m engine outputs, where n and m are positive integers with n greater than m, said method comprising the steps of:
-
(i) providing a computer model of the engine, said model having a plurality of control inputs and having m model outputs; (ii) initializing said multidimensional vector dP to zero; (iii) storing at least one non-square matrix C of sensitivity coefficients which relates changes in performance parameters of the engine dP to a vector of differences between engine sensor outputs and computer model outputs dx, where dx=CdP; (iv) operating said engine; (v) computing said m model outputs based upon said plurality of control inputs and said vector of performance chances dP; (vi) comparing said computed m model outputs to m sensor engine outputs and calculating a vector of output sensor changes dx; (vii) applying singular value decomposition to compute a revised dPe, where C=USS ;
0VT and VR (SS)-1 UT is the pseudoinverse of C. - View Dependent Claims (2, 3, 4, 5, 6)
-
-
7. A method of monitoring changes in an engine, said method specifically monitoring n performance parameters of said engine by means of n performance parameters changes in the form of a multidimensional vector dP, said engine having a plurality of control inputs and resulting in m engine outputs, where n and m are positive integers and n is greater than m, said method comprising the steps of:
-
(i) providing a computer model of the engine, said model having a plurality of control inputs and having m model outputs; (ii) initializing said multidimensional vector dP to zero; (iii) storing at least one non-square matrix C of sensitivity coefficients which relates changes in performance parameters of the engine dP to a vector of differences between engine sensor outputs and computer model outputs dx, where dx=CdP; (iv) operating said engine; (v) computing said m model outputs based upon said plurality of control inputs and said vector of performance changes dP; (vi) comparing said computed m model outputs to m sensor engine outputs and calculating a vector of output sensor changes dx; (vii) applying singular value decomposition to compute a revised dPe, where C=USS ;
0VT and VR (SS)-1 UT is the pseudoinverse of C, by the following steps;(a) operating on the calculated vector of sensor changes dx by (SS)-1 UT to obtain a vector α
;(b) applying a gain factor K to each element of vector α and
integrating the resultant signal with respect to time to obtain vector β
;(c) operating on said vector β
by matrix VR to obtain a calculated vector of performance parameter changes dPe;(viii) substituting said calculated vector of performance changes dPe for the previous vector of performance changes in the computer model in step (ii); and (ix) repeating steps (v) through (viii) until a steady state dPe has been achieved. - View Dependent Claims (8, 9)
-
Specification