Nearly orthogonal latin hypercubes for optimization algorithms

1. Apparatus for improving an initial optimization algorithm that uses a curve-fit small data set and expands the small data set to provide an expanded data set on which curve fitting is imposed to optimize the initial optimization algorithm comprising:

a specialized processor for generating an expanded data set of synthetic data points generated from the set of data points that is the result of creating an initial optimization algorithm using said small data set, the synthetic data points being generated using a nearly orthogonal latin hypercubes algorithm, said expanded data set of synthetic data points being used by said initial optimization algorithm to provide corrected optimization algorithm coefficients.

2. The apparatus of claim 1 , wherein said synthetic data set provided by said nearly orthogonal latin hypercubes algorithm is utilized to modify the initial optimization algorithm to obtain better coefficients of fit.

3. Apparatus for improving an optimization algorithm generated utilizing a small data set that describes a monitored system, comprising:

a small data set generated from a larger data set by considering data only within a predetermined range;

an engine programmed with a nearly orthogonal latin hypercubes algorithm;

a specialized processor for deriving an initial vector corresponding to a curve fitted to the small data set,

said specialized processor utilizing the initial vector to create an initial optimization algorithm, and providing from the initial optimization algorithm an initial data set that fits into said nearly orthogonal latin hypercubes, said specialized processor creating from the nearly orthogonal latin hypercube algorithm a synthetic data set larger than the small set utilized in deriving the initial optimization algorithm;

a curve fitting module for deriving a new vector corresponding to a new curve fitting with the expanded data set; and,

a coefficient correction generator using the new vector to correct coefficients in the initial optimization algorithm.

4. The apparatus of claim 3 , wherein said initial optimization algorithm is created by modeling and simulation utilizing said initial vector.

5. The apparatus of claim 3 , and further including a substitution unit for substituting observed data into the nearly orthogonal latin hypercube algorithm for the initial data set, thereby to create a further expanded synthetic data set corrected by the observed data, such that with new curve fitting to said further expanded synthetic data set, the resulting new curve fitting vector provides predictability that is increased over that associated with the initial optimization algorithm whereby the further expanded data set is used to obtain better coefficient of fit.

6. The apparatus of claim 5 , wherein said further expanded new synthetic data set is corrected for observed data.