METHOD FOR THE COMPUTER-ASSISTED MODELING OF A TECHNICAL SYSTEM

摘要

The invention relates to a method for the computer-assisted modeling of a technical system. At each of a plurality of operating points (t-5, t-4,..., t+3), the technical system is described by a first state vector (it-5, it-4,..., it+3) with one or more first state variables and by a second state vector (ot, ot+1,..., ot+3) with one or more second state variables. A neural network is used for the computer-assisted modeling of said system, said network comprising a special form of a feed-forward network (F1, F2,..., F4). Said feed-forward network (F1, F2,..., F4) is characterized in that at least one connector (K) of the network is a bridging connector that connects a neural layer with the output layer (O) of the feed-forward network, thereby bridging at least one hidden layer (H1, H2, H3). Such a network structure allows the implementation of networks with a plurality of hidden layers that can be trained in a simple manner with known learning methods, in particular the gradient descent method. The method of the invention is used in a preferred variant for modeling a technical system in the form of a gas turbine. The neural network that is trained using the method according to the invention can be used in order to estimate or predict nitrogen oxide or carbon monoxide emissions released by the gas turbine or parameters relating to vibrations of the combustion chamber of the gas turbine.