| WO1990013794A1 | 1990-11-15 |
| US4792737A | 1988-12-20 | |||
| US4847769A | 1989-07-11 | |||
| US4893262A | 1990-01-09 | |||
| US4347730A | 1982-09-07 | |||
| US4928256A | 1990-05-22 | |||
| US4876854A | 1989-10-31 |
| 1. | A method for calibrating readings of a multiple sensor system, the sensors providing output signals in response to external events, the method comprising the steps of: a) providing data base means for storing information on: a plurality of test point sensor output signal values for some of said sensors and a plurality of values for said external events corresponding to said test point sensor output values wherein uncalibrated sensors can be added; said calibrated sensor readings or, alternatively, said readings accompanied with their calibration parameters and values for said external events corresponding to a situation; and, controls of or changes in, if any, said sensors or said external events corresponding to a new situation; b) providing logic means for accessing said calibrated readings or, alternatively, said readings accompanied with their calibration parameters, wherein said logic means may have a twoway communications link to said data base means; c) providing said sensor output signals from said sensors to said logic means; d) providing information, if any, on said controls or changes to said data base means; e) updating by a Kalman recursion wherein the improvement comprises the use of an algorithm obtained from the Fast Kalman Filter (FKF) formulae (28) of the description, in said logic means (1), values of both said external events and said calibration parameters corresponding to said new situation; and, f) providing updated values of said calibrated readings and/or said values of said external events, as desired. |
| 2. | The method of claim 1 wherein said logic means (1) operates in a decentralized or cascaded fashion but exploits in one way or another Kalman filtering wherein the improvement comprises the use of an algorithm obtained from the Fast Kalman Filter (FKF) formulae (28) of the description. |
| 3. | The method of claim 1 including the step of: a) adapting by using Kalman filtering wherein the improvement comprises the use of an algorithm obtained from the Fast Kalman Filter (FKF) formulae (28) of the description, in said logic means (1), said information on said controls of or changes in said sensors or said external events as far as their true magnitudes are unknown. |
| 4. | The method of claim 2 including the step of: a) adapting by using Kalman filtering wherein the improvement comprises the use of an algorithm obtained from the Fast Kalman Filter (FKF) formulae (28) of the description, in said logic means (1), said information on said controls of or changes in said sensors or said external events as far as their true magnitudes are unknown. AMENDED CLAIMS [received by the International Bureau on 29 September 1993 (29.09.93); original claims 14 replaced by amended claim 1 (3 pages)] 1 A method for optimal or suboptimal Fast Kalman Filtering (FKF) where the Measurement and State Equations, in linearized forms, are as follows: yt = Ht st + et (1) s t_. = A r t.l , + B u *.tl + a. (2) where at time t = measurement vector of observed angles, distances, etc. = observations' gain or design matrix st = state vector, calibration parameters included et = measurement error vector R_ = Cov(et) = E e. e' = covariance matrix of said errors τι_ = external forcing or control vector &_ = system noise vector = Cov(at) = E at a' = covariance matrix of said system noise A = state transition matrix which typically depends on time t B = gain matrix of external forcing or controls, and where said vectors and matrices, due to physical or other structural relationships between their elements, can be partitioned as follows: where ct = calibration (typically) parameters for a time/space volume bt ^ = all other state parameters for said volume, and, A. , A. A. and A„ = submatrices of said partitioned A Dl D2 °K c B* , B* ,..., . and BΛ = submatrices of said partitioned B, Dl D2 DK c wherein the improvement comprises the indicated new way of partitioning said state transition matrix A and/or said gain matrix B of said external forcing or said controls, and where the Fast Kalman Filter (FKF) formulae for a recursion step are: where, for k=l,2,...,K, Cov(e k) t ~ Cov{V«trstι)abt , JkJ. and, for k=0, R Λt,o = V v t1,0 where bt ^ and ct are the optimal or suboptimal estimates (Λ) of said state parameters s. and wherein the improvement comprises the possibility to compute their error variances and covariances from: C1+D1SD DjSDJ D1SDK DχS D2SDJ C2+D2SDJ D2SDK D2S DKSDJ DKSD2 CK+DKSDK 'OK? where °k {Xt,kVt!A,k} for k=l,2,...,K D,_ = {x ,k Λk}'lχlkv"lG ** WA.^ and where optimal or suboptimal predictions E ,^ and c t of said state parameters st can be obtained for prediction and process control purposes from: s . = A st_1 + B ut (21) where their error variances and covariances Pt are computed: Pt = Cov( s t) = A CovOs^j) A' + Qt (24) where the " «» "notation indicates that said prediction (no observational data available) step takes place during a time interval (t,tl). STATEMENT UNDER ARTICLE 19 Amended Claim 1 replaces the original Claims 14 that appeared to be too ambigous in specifying the invented generalization of the Fast Kalman Filtering (FKF) method of PCT/FI90/00122 (WO 90/13794, 15.11.90) for use in large dynamical systems that are governed by Partial Differential Equations (PDE). Inspite of this improved formulation of the Amended Claims, the reader may still find it difficult to grasp the scope of this single Amended Claim 1 due to such complexities of the invention that actually belong to Prior Art. It is therefore understood that the description will include a documentation of a PCprogramme that has specifically been prepared for the demonstration purposes without getting unnecessarily involved in the sophistication of large multiparametric dynamical systems. |
Technical Field
This invention relates generally to all practical applications of the Kalman Filter and more particularly to large dynamical systems with a special need for fast, computationally stable and accurate results.
Background Art
Prior to explaining the invention, it will be helpful to first understand the prior art of both the Kalman Filter (KF) and the Fast Kalman
Filter (FKF™) for calibrating a sensor system (WO 90/13794). The underlying
Markov process is described by the equations from (1) to (3). The first equation tells how a measurement vector y t depends on the state vector s t at timepoint t, (t=0,l,2...). This is the linearized Measurement (or observation) equation: y t = H t s t + e t (1)
The design matrix EL is typically composed of the partial derivatives of the actual Measurement equations. The second equation describes the time evolution of e.g. a weather balloon flight and is the System (or state) equation: = -l + u t-l + a t < 2 >
(or, s t = A s t _, + B u t ι + a t more generally) which tells how the balloon position is composed of its previous position s t , as well as of increments IL_, and a t . These increments are typically caused by a known uniform motion and an unknown random acceleration, respectively.
e measurement errors, t e acce era on term an t e prev ous position usually are mutually uncorrelated and are briefly described here by the following covariance matrices:
R e =- Cov(e t ) = E(e t e t ')
R a = Cov(a t ) = E(a t a t ') t and (3)
P t _ 1 =Cov(s t _ =E (s t _ 1 -s t . 1 )(s t _ 1 -s t _ 1 )']-
The Kalman forward recursion formulae give us the best linear unbiased estimates of the present state
^t =ϊ t-ι +11 t.ι +κ t{ y t- H A.ι +,1 t-ι ) } w and its covariance matrix
where the Kalman gain matrix ( is defined by
Let us now partition the estimated state vector s t and its covariance matrix P. as follows:
where b A t. tells us the estimated balloon position; and, c t the estimated calibration parameters.
The respective partitioning of the other quantities will then be as follows:
11 = 0^ :
(8)
The recursion formulae from (4) to (6) gives us now a filtered (based on updated calibration parameters) position vector
and the updated calibration parameter vector
^t-i +u c (10) t-i +κ c t{ < ■y t- H A-ι +u t-ι ) } J
The Kalman gain matrices are respectively
and (11)
The following modified form of the general State equation is introduced
As t-1 +Bu t-1 = l s t +A s t-r s t-l> "a t (12)
where s represents an estimated value of a state vector s. Combine it with the Measurement equation (1) in order to obtain so-called Augmented Model:
The state parameters can now be computed by using the well-known solution of a Regression Analysis problem given below. Use it for Updating:
The result is algebraically equivalent to use of the Kalman Recursions but not numerically. For the balloon tracking problem with a large number sensors with slipping calibration the matrix to be inverted in equations (6) or (11) is larger than that in formula (14).
The initialization of the large Fast Kalman Filter (FKF™) for solving the calibration problem of the balloon tracking sensors is done by Lange's High-pass Filter. It exploits an analytical sparse-matrix inversion formula (Lange, 1988a) for solving regression models with the following so-called Canonical Block-angular matrix structure:
This is a matrix representation of the Measurement equation of an entire windfinding intercomparison experiment or one balloon flight. The vectors b j ,b2,...,b κ typically refer to consecutive position coordinates of a weather balloon but may also contain those calibration parameters that have a significant time or space variation. The vector c refers to the other calibration parameters that are constant over the sampling period.
The Regression Analytical approach of the Fast Kalman Filtering (FKF™) for updating the state parameters including the calibration drifts in particular, is based on the same block-angular matrix structure as in equation (15). The optimal estimates of b j ^,..,!^ and c are obtained by making the following logical insertions into formula (15) for each timepoint t, t=l,2,...:
and, (16)
a c t
These insertions concluded the specification of the Fast Kalman Filter (FKF™) algorithm for calibrating the upper-air wind tracking system. Another application would be the Global Observing System of the World Weather Watch. Here, the vector ^ contains various observed inconsistencies and systematic errors of weather reports (e.g. mean day-night differences of pressure values which should be about zero) from a radiosonde system k or from a homogeneous cluster k of radiosonde stations of a country (Lange, 1988a/b). The calibration drift vector b, will then tell us what is wrong and to what extent. The calibration drift vector c refers to errors of a global nature or which are more or less common to all observing systems (e.g. biases in satellite radiances and in their vertical weighting functions or some atmospheric tide effects).
For all these large multiple sensor systems their design matrices H typically are sparse. Thus, one can usually perform in one way or another the following sort of
Partitioning:
where c. typically represents calibration parameters at time t b t , all other state parameters in the time and/or space volume
A state transition matrix (bock-diagonal') at time t
B matrix (bock-diagonal) for state-independent effects u. at time t.
Consequently, two (or three) types of gigantic Regression Analysis problems
(18)
were faced as follows:
Augmented model for a space volume case: see also equations (15) and (16), e.g. for the data of an entire windtracking experiment with K consecutive balloon positions:
Augmented Model for a moving time, volume (e.g. for "whitening" an observed "innovations" sequence of residuals e. over a moving sample of length L):
Please observe that the matrix formula may take a "nested" Block-Angular structure. Fast semi-analytical solutions based on
ating: s t = {z t > v- i z t } x -1 z.
Upd ;v (i9)
for all these three cases were published in PCT/FI90/00122 (Lange, 1990), WIPO, Geneva, Switzerland.
SUBSTITUTE SHEET
The Fast Kalman Filter (FKF™) formulae for the recursion step at any timepoint t were as follows:
t _ / ={X;_ / V-! / X t . / } "1 X;. / V- 1 y t _ r G t _ / c t ) for /=o,ι,2 L-I
where, for /=0,1,2,...,L-1,
_ -l.
Kr -fc - -'-. y' - 1 -&-f X '--C< l ->
and, i.e. for /=L,
*t-L= V V t- i -L
A major R & D project was initiated in 1988 which led to the start of cooperation between ECMWF and Meteo-France for the coding of a dynamical atmospheric model, an optimal interpolation, a variational data assimilation and a Kalman Filter (FK), all in the same framework. The project is called IFS (Ihtegrated Forecasting System), see Jean-Noel Thepaut and Philippe Courtier (1991): "Four-dimensional variational data assimilation using the adjoint of a multilevel primitive-equation model", Quarterly Journal of the Royal Meteorological Society, Volume 117, pp. 1225-1254.
Similar Kalman Filter (FK) studies have recently been reported also by Roger Daley (1992): "The Lagged Innovation Covariance: A Performance Diagnostic for Atmospheric Data Assimilation", Monthly Weather Review of the American Meteorological Society, Vol. 120, pp. 178-196, and Stephen E. Cohn and David F. Parrish (1991): "The Behavior of Forecast Error Covariances for a Kalman Filter in Two Dimensions", Monthly Weather Review of the American Meteorological Society, Vol. 119, pp. 1757-1785. Unfortunately, the ideal Kalman Filter systems described in the above reports have been out of reach at the present time. Dr. T. Gal-Chen of School of Meteorology, University of Oklahoma, reported in May 1988: "There is hope that the developments of massively parallel super computers (e.g., 1000 desktop CRAYs working in tandem) could result in algorithms much closer to optimal...", see "Report of the Critical Review Panel - Lower Tropospheric Profiling Symposium: Needs and Technologies", Bulletin of the American Meteorological Society, Vol. 71, No. 5, May 1990, page 684.
There exists a need for exploiting the principles of the Fast Kalman Filtering (FKF™) method for a broad technical field (broader than just calibrating a sensor system in some narrow sense of word "calibration") with equal or better computational speed, reliability, accuracy, and cost benefits than other Kalman Filtering methods can do.
Summary of the Invention
These needs are substantially met by provision of the generalized Fast Kalman Filtering (FKF™) method for calibrating and adjusting the sensors and various model parameters of a dynamical system in real-time or in near real-time as described in this specification. Through the use of this method, the computation results include the forecast error covariances that are absolute necessary for warning, decision making and control purposes.
Best Mode for Carrying out the Invention
Prior to explaining the invention, it will be helpful to first understand prior art Kalman Filter (FK) theory exploited in the current experimental Numerical Weather Prediction (NWP) systems. As previously, they make use of equation (1):
Measurement Equation: t = H. s. + e. ...(linearized regression)
where state vector s. describes the state of the atmosphere at timepoint t. Now, s t usually represents all gridpoint values of atmospheric variables e.g. the geopotential heights of a number of different pressure levels.
The dynamics of the atmosphere is governed by a well-known set of partial differential equations ("primitive" equations). Making use of the so-called adjoint operator of the model the following linear expression of equation (2) is obtained for the time evolution of the atmosphere at each time step:
State Equation: s = A s t , + B π t 1 + a t ...(the discre t ized
1 l-1 w l dyn-stoch model)
A
The four-dimensional data assimilation results (s and the NWP forecasts ( sf f ), respectively, are obtained from the Kalman Filter system as follows:
s t = s t + K t (y t - H t S t )
( 21 ) s t " A s t-1 + B u t-1
where P t = Cov( s t ) = A Cov(s^-_ 1 ) A' + Q f .(prediction accuracy)
Q t = Cov(a t ) = E a t a t .(system noise)
R. = Cov(e t ) = E e t e .(measurement noise)
and the crucial Updating computations are based on the following Kalman Recursion:
K t = P t H; (H t P t H i + R t ) ■ 1 ... Kalman Gain matrix) Cov(s t ) = P t - K t H t P t ...(estimation accuracy).
The matrix inversion needed here for the computation of the Kalman Gain matrix is exceedingly difficult for any realistic NWP system because the data assimilation system must be be able to digest several ten thousand data elements at a time.
The method of the invention will now be described. We start with the Augmented Model from equation (13):
A(s t-r s t-ι>- a t
i.e. = Z t s t +
For the four-dimensional data assimilation the following two equations are obtained for its Updating:
or, s t + K t (y t H. s t> ...(alternatively)
and,
Cov(s t ) = E(s t -s t )(s t -s t )' (23)
= |H' R ~ t H t + Pjrl ...(estimation accuracy)
where, as previously, s t - A s^ j + B u, t-l .(NWP "forecasting")
P t = Cov(S t ) = A Cov(s t _ 1 ) A' + Q t (24) but insteaf of κ t « p t HJ (H t p t H; + R 1 .(Kalman Gain matrix)
we take K t = Cov(s t ) HJ R^ 1 (25)
The Augmented Model approach is superior to the use of the Kalman Recursion formulae for a large vector of input data t because the computation A of the Kalman Gain matrix K I. required a huge matrix inversion when Cov(s was unknown. Both methods are algebraically and statistically equivalent but certainly not numerically.
Unfortunately, the Augmented Model formulae above may still become much too difficult to handle numerically if the number of the state parameters is overly large. This actually happens, firstly, if state vector s t consists of enough gridpoint data for a realistic representation of the atmosphere. A spectral decomposition (or empirical orthogonal functions) could be attempted here for the purpose of decreasing the number of state parameters. Secondly, there are many other state parameters that must be included in the state vector for a realistic NWP system. These are first of all related to systematic (calibration) errors of observing systems as well as to the so-called physical parameterization schemes of small scale atmospheric processes.
Fortunately, all these problems are overcome by using the method of decoupling states through exploitation of the general Fast Kalman Filtering (FKF™) method. For the large observing systems of the atmosphere their design matrices H typically are sparse. Thus, one can perform the following
Partitioning:
where c typically represents "calibration" parameters at time t b t , values of atmospheric parameters at grid point k (k=l,...K) A state transition matrix at time t (submatrices A j ,..., g 4 ) B for state-independent effects (submatrices B,,...,B K ,B ).
Consequently, the following gigantic Regression Analysis problem is faced:
The Fast Kalman Filter (FKF ™ ) formulae for the recursion step at any timepoint t are as follows:
where, for k=l,2,...,K,
R = V "1 X t,k{ X t,k V t!k X t,k} X Λ t',k V v " t 1 ,k
The data assimilation accuracies are obtained from equation (23) as follows:
Cov(s t ) = Cov(b t v ... ,ϊ> t κ ,c t ) (29)
where C,. = { X t,k V t!k X t,k} for k=l,2,...,K
Through these semi-analytical means all the matrices to be inverted for the solution of gigantic Regression Analysis models of type shown in equation (27) are kept reasonably small and, especially, for the preferred embodiment of the invention for an operational Numerical Weather Prediction (NWP) model and four-dimensional data assimilation system that is much too large to be specified here. Obviously, the error variances and covariances of the forecasts and the data assimilation results are derived using equations (24) and (29), respectively.
The generalized Fast Kalman Filtering (FKF™) formulae given in equations (28) and (29) are pursuant to the invented method.
Those skilled in the art will appreciate that many variations could be practiced with respect to the above described invention without departing from the spirit of the invention. Therefore, it should be understood that the scope of the invention should not be considered as limited to the specific embodiment described, except in so far as the claims may specifically include such limitations.
References
(1) Kalman, R. E. (1960): "A new approach to linear filtering and prediction problems". Trans. ASME J. of Basic Eng. 82:35-45.
(2) Lange, A. A. (1982): "Multipath propagation of VLF Omega signals". IEEE PLANS '82 - Position Location and Navigation Symposium Record, December 1982, 302-309.
(3) Lange, A. A. (1984): "Integration, calibration and intercomparison of windfinding devices". WMO Instruments and Observing Methods Report No. 15.
(4) Lange, A. A. (1988a): "A high-pass filter for optimum calibration of observing systems with applications". Simulation and Optimization of Large Systems, edited by A. J. Osiadacz, Oxford University Press/Clarendon Press, Oxford, 1988, 311-327.
(5) Lange. A. A. (1988b): "Determination of the radiosonde biases by using satellite radiance measurements". WMO Instruments and Observing Methods Report No. 33, 201-206.
(6) Lange, A. A. (1990): "Apparatus and method for calibrating a sensor system". International Application Published under the Patent Cooperation Treaty (PCT), World Intellectual Property Organization, International Bureau, WO 90/13794, PCT/FI90/00122, 15 November 1990.
Next Patent: FLUIDIC OSCILLATOR AND FLOW METER COMPRISING SAME