Method and apparatus for establishing a data based function model for calculating a direction in a model calculation

摘要

The present invention relates to a method for determining an not parametric, data based function model, in particular of a gauss process model, with is made available by the support points, wherein the supporting points of a number of by one or more input variables defined support locations with respective associated output values of at least one output variable, comprising the steps of:– Selecting (s1) of one or more of the supporting points of support points as a certain;– Formulating (s2) of a system of equations with equations, which each differ by inserting a of the specific support points in the gauss process model equation, wherein the gauss process model equation by hyper parameters (σ$2\; /\; f$, l = (l₁...l_d), σ$2\; /\; n$) and a coefficient vector (q''_i, Q_i) is determined; and– Adapting (s3) of the number of the equations of the equation system and the number of the hyper parameters (σ$2\; /\; f$, l = (l₁...l_d), σ$2\; /\; n$) and components of the vector (q''_i, Q_i), so that the system of equations is detachably;– Release (s4) of the system of equations, about the hyper parameters (σ$2\; /\; f$, l = (l₁...l_d), σ$2\; /\; n$) and the components of the vector (q''_i, Q_i) to obtain.