Assessment of global optimizers: Particle swarm optimization, simulated annealing, and genetic algorithms in local simultaneous multi-loop tuning of PID gains

Shu Zhang, Dong Ye, Joseph Bentsman, Cyrus W. Taft, Aaron Hussey

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Simultaneous multi-loop PID tuning is a procedure for the simultaneous on-line tuning of PID gains in a cluster of PID controllers operating in a closed loop within a multi-variable process. In the previous work [26], a simplified closed-loop system consisting of a multi-input-multi-output (MIMO) nonlinear boiler/turbine model and a nonlinear cluster of six PID-type controllers was specified, with cross-coupling of the variables similar to that in a real power plant, and a local, gradient based, non-model-based method referred to as IFT (iterative feedback tuning) was tested on this system. Based on the figure of merit for the control system performance, the IFT was shown to deliver performance favorably comparable to that attained through empirical tuning by a tuning expert. However, being a gradient-based technique, IFT is capable of finding only local minima potentially located arbitrarily far from the global one. Therefore, it is of interest to investigate performance of model-based, global, non-gradient-based optimizers in simultaneous multi-loop PID tuning. However, before addressing tuning for the global minimum, it is of interest to assess the performance of the global non-gradient-based optimizers in the local simultaneous multi-loop PID tuning. This task is carried out in the present work, comparing the performance of IFT with that of three global optimization techniques: particle swarm optimization (PSO), simulated annealing (SA), and genetic algorithms (GA), using the same model. It is shown that each optimizer is capable of attaining performance comparable to that attained by IFT. The global techniques are compared, and PSO is found to be the least complex, while yielding the tuning performance comparable to that attained by the SA and GA techniques.

Original languageEnglish (US)
Title of host publication53rd ISA POWID Symposium 2010
Pages246-281
Number of pages36
StatePublished - Sep 10 2010
Event53rd ISA POWID Symposium 2010 - Summerlin, NV, United States
Duration: Jun 6 2010Jun 11 2010

Publication series

Name53rd ISA POWID Symposium 2010
Volume483

Other

Other53rd ISA POWID Symposium 2010
CountryUnited States
CitySummerlin, NV
Period6/6/106/11/10

ASJC Scopus subject areas

  • Energy Engineering and Power Technology
  • Nuclear Energy and Engineering
  • Control and Systems Engineering

Fingerprint Dive into the research topics of 'Assessment of global optimizers: Particle swarm optimization, simulated annealing, and genetic algorithms in local simultaneous multi-loop tuning of PID gains'. Together they form a unique fingerprint.

Cite this