METHOD FOR CREATING A MODEL OF A CONTROL SIZE FOR NONLINEAR, STATIONARY REAL SYSTEMS, FOR EXAMPLE, COMBUSTION ENGINES OR SUBSYSTEMS THEREOF

摘要

A method for determining the characteristic diagram of a control variable of an internal combustion engine or a subsystem thereof, preferably using neural networks, which control quantity depends on a set of input variables, for example engine or system parameters and / or default values at the respective operating point, wherein the control variable for a Group of operating points from the entire space of operating points of the engine or subsystem is determined by metrology, at these operating points using a simplified partial model function, an output per model function is determined, and wherein at any operating point, the outputs of each sub-model function with an associated weighting function weighted to a total output for the respective operating point are added together, wherein for all operating points with metrologically determined control variable in each case the difference between the total Ausgangsg size and the metrologically determined value of the control variable is determined and used in areas of operating points with an absolute value of this difference above a predetermined value, a further model function with a further associated weighting function, wherein the absolute value of the difference remains below the predetermined value. In order to obtain faster, e.g. With fewer iterations to arrive at an optimal overall model that satisfies a statistically substantiated high predictive quality and an overall model of as few submodels as possible, the steps of determining the difference between the total output of the associated submodel functions and a real value The control variable and the application of another model and weighting function so often go through until the statistically evaluated predictive quality of the overall model has reached a desired value.