The design of a fast inference and symmetric membership function based fuzzy chip

The fuzzy controller has been used in different kinds of fields such as electrical, mechanical, chemical and aeronautical areas. It is possible to realize this controller using software or hardware. The software implementation has higher cost and takes a longer time to achieve the inference result; while the hardware implementation, fuzzy chip, runs faster and exhibits better compactness at a lower cost. In this paper, we introduce a symmetric-membership-function (SMF) and fast-inference fuzzy chip. The fuzzy chip consists of two inputs and one output. The resolution of input and output membership functions are of six bits, and they are stored in the ROM. The MOH (mean of height) method is used in the defuzzification process. The cell-based ASIC design of this chip uses the technique of the 0.8 μm SPDM CMOS elements. The chip area is 2934 μm×2312 μm, its transistor count equals 6 K approximately, and the maximum clock rate is 62.5 MHz. In summary, we have accomplished the design of a SMF-based and fast-inference fuzzy chip. Its simulation speed can run over 10 M fuzzy rules per second. We also established the design environment for another chip. It is a universal design platform to design another chip thereafter