VLSI implementation of adaptive Fuzzy and neuro controllers

Abstract

newlineThe artificial intelligence techniques such as neural networks and newlinefuzzy systems play an important role in the design of sophisticated control newlinesystems and are used as functional approximators in control applications newlineThese techniques imitate the human expertise based control and are employed newlineto control the complex and or highly nonlinear systems The function of the newlineneural and fuzzy logic controllers is to map the output of the process newlinecorrespond with the control actions in order to maintain the desired newlineperformance of the process The neural networks have been found more newlinesuitable for the control and identification purposes in adaptive control newlineapplications The neural network weights are updated adaptively by tracking newlinethe error online in order to maintain the desired performance of the process newlinedespite variations in the characteristics of the process and disturbances These newlineneural and fuzzy logic controllers need to meet high speed constraints when newlineused for real time applications like autopilots in spacecrafts control of robots newlinespeed control of drives and pH control newlineThe neural networks and fuzzy logic controllers are generally newlineimplemented in software based general purpose computers where the data newlineprocessing is performed sequentially resulting in longer processing time The newlinemassively parallel nature of neural networks when implemented in hardware newlineleads to high speed of operation and in the pure hardware based neural newlinenetworks the execution time is substantially reduced based on the degree of newlineparallelism in the neuron computations data precision and also based on the newlinepipeline architecture newline