Method for computer-supported learning of a control and/or regulation of a technical system

摘要

The preset actions are executed during operation of technical system to transform operation states. Quality function and action-selecting rule are learnt based on training data. A measure of statistical uncertainty of quality function is determined by uncertainty propagation. An optimal operation of technical system modeled by quality function, is determined as function of measure of statistical uncertainty, safety parameter of minimum requirement of quality function and uncertainty propagation using covariance matrix while ignoring non-diagonal elements.