Parallel Implementation of Hybrid MPC
In this chapter parallel implementations of hybrid MPC will be discussed. Different methods for achieving parallelism at different levels of the algorithms will be surveyed. It will be seen that there are many possible ways of obtaining parallelism for hy
- PDF / 270,069 Bytes
- 18 Pages / 439.37 x 666.142 pts Page_size
- 30 Downloads / 204 Views
Parallel Implementation of Hybrid MPC D. Axehill and A. Hansson
Abstract In this chapter parallel implementations of hybrid MPC will be discussed. Different methods for achieving parallelism at different levels of the algorithms will be surveyed. It will be seen that there are many possible ways of obtaining parallelism for hybrid MPC, and it is by no means clear which possibilities that should be utilized to achieve the best possible performance. To answer this question is a challenge for future research.
23.1 Introduction Speed in numerical computations has increased dramatically for a long period of time. This is partly due to increase of processor performance in computers and partly due to development of more sophisticated algorithms and methods. However, for the last five years single-core processor performance has not significantly increased. To compensate for this multi-core and multi-processor computers have seen an increased use. In addition to this clusters and grids have emerged as another way to speed up computations. Multi-core and multi-processor computers typically have only few cores and processors whereas clusters and grids can be composed of hundreds of processors distributed over a significant number of computers. It is clear that these
With kind permission from Springer Science+Business Media: Distributed Decision Making and Control, Towards Parallel Implementation of Hybrid MPC—A Survey and Directions for Future Research, 417/2012, 2012, 313–338, D. Axehill and A. Hansson, figure 14.2, © SpringerVerlag London Limited 2012.. D. Axehill (B) · A. Hansson Division of Automatic Control, Linköping University, Linköping, Sweden e-mail: [email protected] A. Hansson e-mail: [email protected]
J. M. Maestre and R. R. Negenborn (eds.), Distributed Model Predictive Control 375 Made Easy, Intelligent Systems, Control and Automation: Science and Engineering 69, DOI: 10.1007/978-94-007-7006-5_23, © Springer Science+Business Media Dordrecht 2014
376
D. Axehill and A. Hansson
new architectures pose new challenges on how algorithms for numerical computations should be designed. In case care is not taken, potential speedup will not happen. Model Predictive Control (MPC) is a popular control strategy which has been used in many applications for a long time. In later years work has been carried out to generalize MPC to so-called hybrid systems [16]. For these systems the computational demand is even higher. Hybrid systems have applications in, e.g., transportation, logistics, economics, process control, building control, airplane routing and communications. In recent years there has been a trend in the control community to develop distributed algorithms for control. This type of distributed control has much in common with parallel implementations and a parallel implementation of an optimization algorithm for MPC can be interpreted as a form of a distributed MPC controller. However, it should be stressed that the main objective in this work is computational performance. One of the few references available f
Data Loading...
