Abstract
For simulation and control of a district heating system it is of interest to build a mathematical model of the distribution network. This paper deals with stochastic models of the time-variation of temperatures in distribution networks. Due to variations of the water flow in the distribution network it is not possible to apply traditional modeling techniques assuming a constant time-delay between temperature signals. Two methods for tracking both the delay and the dynamics of an input-output system are presented. The methods are based on recursive estimation of the parameters of ARMAX models. In the first method simultaneous recursive estimation of the parameters of a number of ARMAX models with different delays is performed. The second method tracks the states - amongst these the delay - of a state space model formulation. The performance of the estimation methods is illustrated on data from a district heating system. It appears that the performance of the estimation methods depends strongly on the dynamical characteristics of the variations of the delay. The best results are achieved for slowly varying delays. As both methods are based on adaptive schemes they are well suited for on-line applications. Furthermore, the generality of the methods make them applicable
More>>
Citation Formats
Soegaard, H T, and Madsen, H.
On-line estimation of time-varying delays in district heating systems.
Denmark: N. p.,
1991.
Web.
Soegaard, H T, & Madsen, H.
On-line estimation of time-varying delays in district heating systems.
Denmark.
Soegaard, H T, and Madsen, H.
1991.
"On-line estimation of time-varying delays in district heating systems."
Denmark.
@misc{etde_10113399,
title = {On-line estimation of time-varying delays in district heating systems}
author = {Soegaard, H T, and Madsen, H}
abstractNote = {For simulation and control of a district heating system it is of interest to build a mathematical model of the distribution network. This paper deals with stochastic models of the time-variation of temperatures in distribution networks. Due to variations of the water flow in the distribution network it is not possible to apply traditional modeling techniques assuming a constant time-delay between temperature signals. Two methods for tracking both the delay and the dynamics of an input-output system are presented. The methods are based on recursive estimation of the parameters of ARMAX models. In the first method simultaneous recursive estimation of the parameters of a number of ARMAX models with different delays is performed. The second method tracks the states - amongst these the delay - of a state space model formulation. The performance of the estimation methods is illustrated on data from a district heating system. It appears that the performance of the estimation methods depends strongly on the dynamical characteristics of the variations of the delay. The best results are achieved for slowly varying delays. As both methods are based on adaptive schemes they are well suited for on-line applications. Furthermore, the generality of the methods make them applicable to a wide class dynamic systems which contains time-varying delays. (au).}
place = {Denmark}
year = {1991}
month = {May}
}
title = {On-line estimation of time-varying delays in district heating systems}
author = {Soegaard, H T, and Madsen, H}
abstractNote = {For simulation and control of a district heating system it is of interest to build a mathematical model of the distribution network. This paper deals with stochastic models of the time-variation of temperatures in distribution networks. Due to variations of the water flow in the distribution network it is not possible to apply traditional modeling techniques assuming a constant time-delay between temperature signals. Two methods for tracking both the delay and the dynamics of an input-output system are presented. The methods are based on recursive estimation of the parameters of ARMAX models. In the first method simultaneous recursive estimation of the parameters of a number of ARMAX models with different delays is performed. The second method tracks the states - amongst these the delay - of a state space model formulation. The performance of the estimation methods is illustrated on data from a district heating system. It appears that the performance of the estimation methods depends strongly on the dynamical characteristics of the variations of the delay. The best results are achieved for slowly varying delays. As both methods are based on adaptive schemes they are well suited for on-line applications. Furthermore, the generality of the methods make them applicable to a wide class dynamic systems which contains time-varying delays. (au).}
place = {Denmark}
year = {1991}
month = {May}
}