Inference method and parallel implementation for MISO structure systems for inputs with linguistic values

In fuzzy modeling, both clear and fuzzy values can be given to the inputs of the simulated systems. The computational complexity of fuzzy inference with fuzzy inputs, which are a formalization of linguistic values, corresponds to exponential complexity. This paper describes a new method of inference based on the decomposition theorem of multidimensional fuzzy implication and fuzzy truth value. This method makes it possible for fuzzy inputs to implement an inference with polynomial computational complexity, which makes it effective for modeling large-dimensional MISO structure systems. The implementation of this method using parallel computing technologies is reviewed in detail. As a result of the experiment, conclusions were made about the feasibility of using a particular implementation, depending on the amount of input data.