FUZZYSTUDIO™ 2.0 Hardware and Software Support for VLSI Fuzzy Engine

SGS-THOMSON Microelectronics, one of the leaders in the emerging world of fuzzy processing, has introduced a development system for its low cost Weight Associative Rule Processor (W.A.R.P. 2.0), the world’s first 8-bit VLSI fuzzy logic Co-Processor. Known as FUZZYSTUDIO™ 2.0, the kit consists of an Application Development Board that is connected to a standard PC (386 or higher) via an RS-232 port and a Windows-based Software Development Tool that allows the designer to program the W.A.R.P. 2.0 processor using intuitive graphical and linguistic methods.

The development board contains a W.A.R.P. 2.0 chip, a ZEROPOWER memory to store the user's rules and membership functions and an ZEROPOWER programmer controlled by the host PC. The board is a powerful tool for developing, debugging and testing the hardware implementation of a fuzzy project directly on the physical system to be controlled.

The FUZZYSTUDIO™ 2.0 software tools, which include an Editor, a Compiler, a Debugger and an Exporter, were specially developed to present a user-friendly interface. For example, the Editor allows the user to build a project by using the typical terminology and objects of Fuzzy Logic. Input and Output Variables and Membership Functions can be graphically defined in the Variable Editor, where a user-friendly graphical interface makes it easy to perform the editing phase of the project. Fuzzy Rules can be easily defined by means of linguistic terms in the Rule Editor. The designer is guided in the definition of the rules so to prevent possible syntax mistakes.

The Compiler compiles the application developed by the user and generates the object code to be loaded in the Weight Associative Rule Processor. The Debugger allows the user to step through the fuzzy project one rule at a time by using a software model of W.A.R.P.2.0, while the Exporter generates a W.A.R.P.2.0 model for use in MATLAB, in C programs or in the W.A.R.P.2.0 Simulator environment.

FUZZYSTUDIO™ 2.0 also includes the Adaptive Fuzzy Modeller (AFM), an advanced tool that automatically generates a database containing inference rules and membership function parameters from a given sampling of input and output variables. Using techniques based on neural networks, the AFM allows engineers with no expertise in fuzzy logic to convert their high level systems expertise directly into a fuzzy logic implementation.

January 1997

Top