Dynamic GMDH neural networks and their application in fault detection systems

In this paper, the problem of the dynamic GMDH (Group Method and Data Handling) neural networks and their application in fault detection systems is presented. Such networks can be considered as feedforward networks with a growing structure during the training process. The GMDH networks application in fault detection systems improves their efficiency with lack of information regarding the structure and dynamics of the diagnosed system. The proposed networks have been implemented in fault detection systems using the real data from the Lublin sugar factory.