Development of Explicit Neural Predictive Control Algorithm Using Particle Swarm Optimisation
- 1.3k Downloads
This paper describes development of a nonlinear Model Predictive Control (MPC) algorithm. The algorithm is very computationally efficient because for control signal calculation an explicit control law is used, no on-line optimisation is necessary. The control law is implemented by a neural network which is trained off-line by means of a particle swarm optimisation algorithm. Inefficiency of a classical gradient-based training algorithm is demonstrated for the polymerisation reactor. Moreover, the discussed MPC algorithm is compared in terms of accuracy and computational complexity with two suboptimal MPC algorithms with model linearisation and MPC with full nonlinear optimisation.
KeywordsProcess control Model Predictive Control neural networks optimisation particle swarm optimisation soft computing
Unable to display preview. Download preview PDF.
- 1.Eberhart, R.C., Shi, Y., Kennedy, J.: Swarm Intelligence. Morgan Kaufmann (2001)Google Scholar
- 8.Maciejowski, J.M.: Predictive control with constraints. Prentice Hall, Harlow (2002)Google Scholar
- 12.Pillay, N., Govender, P.: Particle swarm optimization of PID tuning paremeters: optimal tuning of single-input-single-output control loops. LAP Lambert Academic Publishing (2010)Google Scholar
- 16.Yousuf, M.S.: Nonlinear predictive control using particle swarm optimization: application to power systems. VDM Verlag (2010)Google Scholar