Data driven neuroendocrine pid controller for mimo plants based adaptive safe experimentation dynamics algorithm

Mohd Riduwan, Ghazali (2020) Data driven neuroendocrine pid controller for mimo plants based adaptive safe experimentation dynamics algorithm. PhD thesis, Universiti Malaysia Pahang (Contributors, Thesis advisor: Ahmad, Mohd Ashraf).

Data driven neuroendocrine pid controller for mimo plants based adaptive.wm.pdf

Download (2MB) | Preview


This study focused on data-driven tools and controller structure in the data-driven control scheme. Data-driven tools are an optimization method to find the optimal controller parameters using the system’s input and output data. Meanwhile, the controller structure refers to the controller design that is highly dependent on the input and output system. The existing data-driven neuroendocrine-PID (NEPID) utilizes the simultaneous perturbation stochastic approximation (SPSA) algorithm as the data-driven tool. However, this SPSA-based method is unable to find the optimal value of the design parameter due to unstable convergence obtained that degrades the controller performance in MIMO systems. Thus, a safe experimentation dynamics (SED) algorithm is selected to solve this unstable convergence but still not enough to achieve high accuracy because the update designed parameter only depends on the fixed step size gain. For the controller structure, the hormone secretion rate parameter of the existing NEPID is constant during the experimental time. However, control accuracy is insufficient because the secretion rate and control variable error are not able to interact directly and limits the controller capability. Besides, in the existing NEPID controller structure of the SISO system, only a single node of hormone regulation is used due to a single control variable. Meanwhile, in the MIMO systems, many control variables available that interact with each other, and the single node hormone regulation of NEPID is still inadequate in producing better control accuracy of nonlinear MIMO systems. Therefore, this study proposed the adaptive safe experimentation dynamics (ASED) algorithm to improve the SED algorithm performance accuracy by minimizing its objective function in terms of mean, best, worst, and standard deviation analysis. In order to increase the control accuracy of the existing NEPID controller, this study also established the sigmoid-based secretion rate neuroendocrine- PID (SbSR-NEPID) controller structure by varying the hormone secretion rate according to the change of error. Finally, this study also focused on developing a multiple node hormone regulation neuroendocrine-PID (MnHR–NEPID) controller structure to improve the control accuracy of existing NEPID by prioritizing the control regulation of each node from their level of error. The performance of PID and NEPID controllers was compared with those of SbSR-NEPID and MnHR-NEPID performances based on error and input tracking. The results show that the ASED- and SED-based methods produced stable convergence. The ASED-based method provided better tracking performance than the SED method by obtaining the objective function’s lower values. Besides, from the simulation work, the SbSR-NEPID and MnHR-NEPID designs provided better control accuracy in terms of lower objective function, total norm of error, and total norm of input compared to those of the PID and NEPID controllers. Moreover, the SbSR-NEPID controller achieved control accuracy improvement of 4.95% and 5.89% for the container gantry crane and TRMS systems, respectively. Besides, the MnHR-NEPID controller achieved control accuracy improvement of 5.7% and 5.1% for the container gantry crane and TRMS systems, respectively. The ASED-based method significantly improved the SED method’s accuracy by using adaptive terms based on changing the objective function in the updated procedure. Besides, the SbSR-NEPID was effective in reducing the error in a transient state, and MnHR-NEPID provided effective interaction between nodes available in MIMO systems which contributed to accuracy improvement.

Item Type: Thesis (PhD)
Additional Information: Thesis (Doctor of Philosophy) -- Universiti Malaysia Pahang – 2021, SV: ASSOCIATE PROF. DR. MOHD ASHRAF BIN AHMAD, NO. CD: 12888
Uncontrolled Keywords: Neuroendocrine-PID (NEPID): simultaneous perturbation stochastic approximation (SPSA)
Subjects: T Technology > TK Electrical engineering. Electronics Nuclear engineering
Faculty/Division: Faculty of Electrical and Electronic Engineering Technology
Depositing User: Mrs. Neng Sury Sulaiman
Date Deposited: 13 Apr 2022 06:47
Last Modified: 31 May 2023 06:38
Download Statistic: View Download Statistics

Actions (login required)

View Item View Item