The goal is the development of efficient off-line control strategies to generate patient-individually optimized process controls for the dialysis therapy. Thus a dynamic optimization problem based on a multicompartment model describing the relevant exchange processes during dialysis is formulated and embedded in a two-loop control concept. In this context the non-applicability of standard methods for the posed problem is discussed. In order to construct efficient control algorithms the dynamic optimization problem is transformed to a pseudodynamic parameter optimization problem (pdPOP). A first approach to the off-line control problem is based on the Bellman-principle of optimality and yields a Step-by-Step algorithm. However, the central point is the development of a much more efficient control algorithm named PSEUDYGALG. This method can be characterized as an iterative and numerical descent method and can be applied to control problems of a variety of science disciplines, if these problems are fulfilling according conditions. High efficiency of the algorithm is achieved by its adaption to the formulated pdPOP and by the use of structural characteristics of the underlying model for the development of update procedures. This efficiency principally renders possible the implementation of this algorithm within an on-line control concept.
|Translated title of the contribution||Optimal control-strategies for nonlinear, physiological systems applied to the hemodialysis process-optimization|
|Qualification||Doctor of Technology|
|Publication status||Published - 20 May 1997|
- descent method
- control problem