Self-organizing fuzzy logic control for real-time processes

This study describes the application of the self-organizing fuzzy logic controller (SOFLC) proposed by S. Daley and K.F. Gill (1986) to real-time control of small scale pilot processes, both single-variable and multivariable. The controller can generate and modify control rules to improve its performance. It is concluded that SOFLC is capable of dealing well with a noise-contaminated environment, and for a slow process the computational burden can be accommodated on a low-cost PC. The multivariable process employed has short time constants and necessitates a sampling period of 10-20 msec. A conventional PC is inadequate for performing the SOFLC and associated functions at such a sampling rate. Thus, a transputer-augmented PC system with a single transputer implementation is used for the implementation. The selection of the scaling factors is discussed.<>