GMDH is an algorithm which can train and produce predicative models following the polynomial neural network(PNN) architecture. The advantage of the algorithm is that it can automatically select the number of inputs and the topology for which accurate results will be plausible,The algorithm utilizes a fitness function to identify the optimal configuration of the neural networks which can generate reliable results.
Problems in wave energy and its converters have never used GMDH techniques to solve or mitigate the uncertainty faced in such systems. Although some studies are there regarding location selection for wave energy systems, but there is still an ample scope to implement GMDH for finding solution to common or uncommon problems faced by the wave energy converters.
You may find some ideas from UO in this aspect.
That's all for now.
