The Time Series Analysis Group
Research
Examples of developments and application in energy systems,
finance, environmental
systems and chemometrics:
Predictions of
electrical
consumption needs
Predictive regulators
for
electrical powerplants
Predictions of
electrical
consumption needs
Predictions of energy
production
in windmill parks
Predictions of interest
rates
On-line control systems
Experimental designs
The models we have developed, include models for:
The oxygen content in
riverwater
Processes in waste water
sewage
plants
Building dynamics
Fuelling dynamics for
car
engines
To develop these applications we use methods from a range of different
disciplines of which the most important are linear and non-linear time
series analysis in discrete and continuous time, as well as linear and
non-linear regression and non-parametric methods.
Among the more theoretical projects concerning methods to model
stochastic,
dynamical systems, we wish to mention the so-called grey-box modelling,
which combine the deductive (deterministic) and inductive (stochastic)
methods to model dynamical systems. Specifically, the Time Series group
have developed methods and programs for identification and estimation
of
grey-box models.
Of other models we apply is Markov Chain models, ARIMA models,
diffusion
models, neural network and state space models, as well as a large range
of non-linear and non-stationary models should be mentioned.
In connection to these models, investigation into and developments
of model identification, parameter estimation and model validation
methods
are researched.
Back to Top,
The Time Series homepage.
The Section for Statistics.
Last modified 23 November 2004. Comments and suggestions to jkl@imm.dtu.dk