A fuzzy neural network approach to model hydraulic component from input/output data

The dynamics of hydraulic components are vital for the virtual prototyping of fluid power systems. This paper proposes a fuzzy neural network approach to model the behavior of a hydraulic component from its input-output. The main advantage of this approach is that the network structure can be determined based on the analysis of the input variables to output response, without trial and error, network pruning or network growing techniques. The process involves resolving the significant inputs through an analysis of their effects with respect to the output. The number of fuzzy rules is determined based on partitioning the input-out space. The number of significant inputs and the number of fuzzy rules together defined the fuzzy neural network structure. A hydraulic pressure relief valve is used to demonstrate the proposed approach. The results indicate that the structure of the fuzzy neural network determined based on the proposed approach can effectively model the dynamics of the relief valve. This work con stitutes initial effort towards determining the structure of neural networks based on the analysis of input-output data.