Pro-Two: a hardware based platform for real time type-2 fuzzy inference

Abstract

This paper describes Pro-Two, a hardware-based platform for implementing two input-one output interval type-2 fuzzy systems. First, the paper describes a computational model for interval type-2 fuzzy systems that considers parallel inference processing and inner-outer bound sets-based type reduction. Then, taking into account this model, we present our architecture for hardware implementation with several pipeline stages for full parallel execution of fuzzy inferences. Such architecture is used to specify Pro-Two, which is implemented over FPGA technology. Experimental results show that Pro-Two performs more than 30 millions of type-2 fuzzy inferences per second for a nine-rule system. We validate the correct functioning of Pro-Two by implementing on it a type-2 fuzzy adaptive filter. The results obtained corroborate the superior performance of type-2 over type-1 fuzzy logic for this application and its increased tolerance to low arithmetic resolution.