Novel Neural Network Modeling Method and Applications

摘要

Neural networks play an important role for designing the parametric modelrnof electromagnetic structures. The current neural network methods are unfit for a circuitrnmodel with many input variables because it is costly to extract a large number of the trainingrndata and test data to complete the highly nonlinear mapping approximation. This articlernproposes a new neural network modeling method—the multidimensional neural networkrnmodel, which can be used to solve the issue of multivariable radiofrequency and microwavernpassive device modeling. The entire multidimensional neural network modeling problem isrnsimplified into a set of neural network submodels through decomposition method. Then thernsubmodels are combined into an equivalent model, and the final entire model is producedrnthrough the neural-network mapping model developed with the submodels and equivalentrnmodel. A microstrip hairpin filter model is developed using the proposed method. The simulationrnresults show the correctness and the effectivity of the proposed method.