Non-linear model reduction by genetic algorithms with using a system structure related fitness function

Based on a known order reduction method for non-linear systems a solution is proposed to reduce the high system complexity of the order-reduced system, too. For this, suitable secondary conditions for the order reduction method are defined with the help of a genetic algorithm (GA). For the use of GA it is essential that the fitness function fulfils some "smoothness" or "small causes, small effects" properties. This is investigated for a system structure related fitness function and an example with technical background is given.