Data-Driven Neuroendocrine-PID Tuning Based on Safe Experimentation Dynamics for Control of TITO Coupled Tank System with Stochastic Input Delay

Abstract

This paper addresses a data-driven neuroendocrine-PID tuning for control a two-input-two-output (TITO) coupled tank system with stochastic input time delay based on safe experimentation dynamics (SED). The SED algorithm is an optimization method used as data-driven tools to find the optimal control parameters by using the input-output (I/O) data measurement in an actual system. The advantages of the SED algorithm are that provides a fast solution, able to solve the high dimensional problem and provides high-performance accuracy by keeping the best parameter value while finding the control parameters. Moreover, the gain sequences of the SED algorithm is independent of the number of iterations by fixed the interval size in finding the optimal solution. Hence, this allows the SED method to have enough strength to re-tune in the attempted of finding the new optimal solution when the delay occurs during the tuning process. Apart from that, a neuroendocrine-PID controller structure is chosen due to its provide effective and accurate control performances by a combination of PID and neuroendocrine structures. On another note, the neuroendocrine structure is a biologically inspired designed that derived from general secretion rules of the hormone in the human body. In order to evaluate the performances of the data-driven neuroendocrine-PID control based on SED, it is applied to a numerical example of TITO coupled tank plant and the control performance tracking and the computational time are observed. The simulation results show that the data-driven neuroendocrine-PID control based on SED capable to track the desired value of liquid tanks level although the stochastic input delay occurred in the system. In addition, the SED based method also attained good control performance without any theoretical assumptions about plant modelling.