Analog synthesis of nonlinear functions based on fuzzy logic

A circuit for the analog implementation of nonlinear functions has been designed. It is based on the fuzzy-logic paradigm mapped into a modular architecture. The basic building blocks rely on the large-signal characteristic of MOS differential pairs to design bell-shaped basis functions and combine them with the center-of-gravity method. The actual input/output (I/O) characteristics are defined by applying proper voltage levels to the high-impedance input nodes of each module. Both a field-programmable processor, where the programming configuration is stored in a digital RAM, and a tool for generating the layout of dedicated circuits, where this connection is hardwired, have been designed using a CMOS 0.7 /spl mu/m n-well technology. A software layer computes the programming values which allow the circuit response to approximate the target I/O relationship. Experiments show that the approximation accuracy is in the range of a few percent r.m.s. of the circuit output range. The application of the system to the synthesis of a nonlinear control law for a DC-DC power converter is discussed.