A Kind of WorldFIP Distributed Intelligent Measurement and Control Network Based on Single Neuron Predictive Identification Control Algorithm

Abstract

Traditional PID control can not meet the control requirement of time-varying process with transport delay. Neural network related algorithms are difficult to be implemented in distributed measurement and control based on fieldbus. A kind of adaptive predicative identification control algorithm based on single neuron was proposed in this paper. A single neuron was used to implement the dynamic identification of controlled object. Tapped delay links (TDL) were used to input information to neuron. Former information was used to correct the predicted output values of predictor to improve prediction precision. Another neuron was used to implement intelligent adaptive control. Three inputs were constructed similarly to PID control. Gradient descent algorithm was used in modifying weights in the neuron. WorldFIP distributed intelligent measurement and control network based on the proposed algorithm was designed. Architecture and construction of the designed system were expounded. Hardware and software design for measurement node, control node and supervisory node were presented. Simulation research and site practice were done. The proposed SNPIC algorithm and designed control network were easier to be implemented and showed better control effects.