A new scheme for an automatic generation of multi-variable fuzzy systems

We present a new novel method of automatically generating a multi-variable fuzzy inference system from given sample sets. We rst decompose the sample set, say , into a cluster of sample sets associated with the given input variables, then compute the associated fuzzy rules and membership functions for each variable, independent of the other variables, by solving a single input multiple outputs fuzzy system extracted from the set cluster. The resulting decomposed fuzzy rules and membership functions for all the variables are integrated back into the fuzzy system appropriate for the original sample set. Taking an advantage of the independence of the input variables in computing the decomposed systems, we show that the computational complexity of the multi-variable system can in principle be reduced to that of a single variable if we can use a parallel processing multi-CPUs system. We have veriied our claim using an eight variable nonlinear function.