Thesis for the Degree of Doctor of Technology


Year 2011
Nikolai Vatanski

Networked control with delay measurement and estimation


As future industrial communication network solutions move towards industrial Ethernet based solutions, monitoring and control of network parameters and control of the network will gain more attention than previously in order to achieve unperturbed performance. The main challenges in implementation of Ethernet based solutions in control applications come from the fact that the performance of the network is priory nondeterministic and may change according to the load in the network. Respectively, this perturbed operation of the communication network may affect on efficiency and performance of the control application.

This work focuses on performance improvement of a control application that is controlled over a switched Ethernet network. In such networks main network parameter that may pose perturbation in performance is the network-induced delay. With novel network instrumentation and network performance estimation methods the behavior of the network-induced delay in the switched Ethernet network is described. The outputs of these methods are used in the control strategies to compensate for the deterious effect of the network.

First, a method for measuring the network information, particularly the network-induced transmission delay, is presented. Next, this information is utilized in control design. Two control approaches are developed, which differ with respect to the time-variant behavior of the delay measurements. A Smith predictor based approach is proposed for the case where network time delay variation is low, and polynomial system based time variant controller and observer design is proposed for the case where time variation is high. Second, a methods for estimating the network-induced transmission delay, is presented and utilized in control design. Two control approaches based on robust control theory are developed, which differ with respect to ability to take into account in addition of delay variation also other uncertainty sources, such as model uncertainty and measurement errors.

The developed control designs are tested with simulations as well as on the experimental platform. From the results it is clear that utilization of the methods improve performance of the application. From the results it is also clear that no single method is suitable for all the cases, but the choice should be based on detail investigation of available instrumentation, magnitude of network delay variation, and uncertainty sources coming from the process.

Thesis electronical version can be downloaded from here

This info last modified 23 Sep 2017 by Jerri Kämpe-Hellenius