Loading...
Development of methods for parallel computation of the solution of the problem for optimal control
Author(s)
Mbangeni, Litha
Date Issued
2010
Type
Thesis
Publisher
Cape Peninsula University of Technology
Abstract
Optimal control of fermentation processes is necessary for better behaviour of the
process in order to achieve maximum production of product and biomass. The problem
for optimal control is a very complex nonlinear, dynamic problem requiring long time for
calculation Application of decomposition-coordinating methods for the solution of this
type of problems simplifies the solution if it is implemented in a parallel way in a cluster of
computers. Parallel computing can reduce tremendously the time of calculation through
process of distribution and parallelization of the computation algorithm. These processes
can be achieved in different ways using the characteristics of the problem for optimal
control.
Problem for optimal control of a fed-batch, batch and continuous fermentation processes
for production of biomass and product are formulated. The problems are based on a
criterion for maximum production of biomass at the end of the fermentation process for
the fed-batch process, maximum production of metabolite at the end of the fermentation
for the batch fermentation process and minimum time for achieving steady state
fermentor behavior for the continuous process and on unstructured mass balance
biological models incorporating in the kinetic coefficients, the physiochemical variables
considered as control inputs. An augmented functional of Lagrange is applied and its
decomposition in time domain is used with a new coordinating vector. Parallel computing
in a Matlab cluster is used to solve the above optimal control problems. The calculations
and tasks allocation to the cluster workers are based on a shared memory architecture.
Real-time control implementation of calculation algorithms using a cluster of computers
allows quick and simpler solutions to the optimal control problems.
process in order to achieve maximum production of product and biomass. The problem
for optimal control is a very complex nonlinear, dynamic problem requiring long time for
calculation Application of decomposition-coordinating methods for the solution of this
type of problems simplifies the solution if it is implemented in a parallel way in a cluster of
computers. Parallel computing can reduce tremendously the time of calculation through
process of distribution and parallelization of the computation algorithm. These processes
can be achieved in different ways using the characteristics of the problem for optimal
control.
Problem for optimal control of a fed-batch, batch and continuous fermentation processes
for production of biomass and product are formulated. The problems are based on a
criterion for maximum production of biomass at the end of the fermentation process for
the fed-batch process, maximum production of metabolite at the end of the fermentation
for the batch fermentation process and minimum time for achieving steady state
fermentor behavior for the continuous process and on unstructured mass balance
biological models incorporating in the kinetic coefficients, the physiochemical variables
considered as control inputs. An augmented functional of Lagrange is applied and its
decomposition in time domain is used with a new coordinating vector. Parallel computing
in a Matlab cluster is used to solve the above optimal control problems. The calculations
and tasks allocation to the cluster workers are based on a shared memory architecture.
Real-time control implementation of calculation algorithms using a cluster of computers
allows quick and simpler solutions to the optimal control problems.
Additional information
Thesis (MTech(Electrical Engineering))--Cape Peninsula University of Technology, 2010
File(s)![Thumbnail Image]()
Loading...
Name
Development of methods for parallel computation of the solution of the problem for optimal control.pdf
Size
6.54 MB
Format
Adobe PDF
Checksum
(MD5):b74014959ac4253a8422f931ce14d77c