Beobachtergestützte fehlertolerante Überwachung drahtlos vernetzter regelungstechnischer Systeme
Final Report Abstract
The main goal of this project is the codesign and joint optimization of the control and the communication subsystems of a wireless networked control system (WNCS), which is in contrast to the traditional design approach that separates control and communication. The observer-based fault-tolerant monitoring system is designed by taking into account the real-time capability and reliability of the communication system. Several measures have been considered to adapt the communication system to the needs of the control system: An upper bound on packet transmission delays is a general requirement of the control application. Under this constraint, the communication system has to work as reliable as possible. However, the delay, packet loss and transmission errors can still influence the performance of the monitoring system. Therefore, the parameters of the communication system, i.e. upper bound of the delay, knowledge of packet loss, and mean square error (MSE) of the analog measurements, are first modeled from the point of view of control and then integrated into the design of observerbased monitoring systems. In this way, the obtained monitoring system is robust against the transmission delay, packet loss and transmission errors, and at the same time sensitive to faults. The whole design procedure is developed in the framework of Markov jumping linear systems (MJLS) with uncertainties. The developed monitoring system adapts to the quality of the communication channels. From the communication point of view, it has to be pointed out that the most important properties of a communication system for control purposes are the real-time capability as well as a high transmission reliability. The real-time capability was achieved by assuring an upper bound for the transmission delay by employing a shared medium access method based on time division multiple access (TDMA). Essentially, each control period is divided into time slots that are assigned to transmissions of certain nodes, according to a dynamic communication schedule. One node has the role of a master node that needs to keep the clocks of all other nodes synchronized, and also to provide at run-time the communication schedule. The logical choice for such a master node is the controller that can actually be seen as a central coordinator of the networked control system. We have concentrated on two strategies of making the communication more reliable. The first strategy is to use the remaining free time slots in a system cycle for flexible retransmissions of lost packets. The second employed strategy is the use of a source/channel coding scheme for the analog-to-digital (AD) converted measurements that allows for reconstructing as much information as possible in case of transmission errors. In a WNCS all transmissions serve to transmit physical values, on the one hand measurements from the sensors to the controller and on the other hand control commands from the controller to the actuators. Therefore, the use of a quantizer with natural binary code (NBC) labeling is proposed and it is shown that a certain class of linear unequal error protection (LUEP) block codes, obtained from low rate, high memory convolutional codes by direct truncation (DT) termination, matches the source characteristics of this kind of quantizer and leads to a low MSE. The unequal error protection (UEP) capabilities of such codes are examined and simulation results are presented that show that these codes minimize the MSE of the analog physical values for a binary symmetric channel (BSC). Additionally, a code construction method adopted from the concept of time-varying convolutional codes was developed that allows for designing MSE minimizing codes that can be adapted to all channel conditions. A real-time experimentation platform named WiNC has been built in order to apply and test the developed concepts. It was realized with off-the-shelf hardware and open source software. The platform is composed essentially of a control application and a driver of a standard Wireless Local Area Network (WLAN) interface card which contains a protocol stack optimized for industrial wireless communication. As benchmark two laboratory setups were considered, a Three-Tank and an Inverted Pendulum, which are regarded as subsystems belonging to a larger system that will be remotely controlled via the wireless network. The first subsystem is composed of three level sensors and two pumps as actuators. The second subsystem has two sensors measuring cart position and rod angle, respectively, and one actuator represented by a servomotor moving the cart. Each sensor and actuator constitutes an independent wireless node, and so the whole system comprises eight nodes that are controlled and coordinated by a remote controller node.
Publications
- Fault detection over noisy channels. 46th IEEE Conference on Decision and Control, New Orleans, USA, 2007
W. Li, P. Zhang, S. X. Ding and O. Bredtmann
- A new fault detection scheme for networked control systems subject to uncertain time-varying delay. IEEE Trans. on Signal Processing, 56, 5258-5268., 2008
Y. Wang, S. X. Ding, H. Ye und G. Wang
- Advanced Design Scheme for Fault Tolerant Distributed Networked Control Systems. IFAC World Congress, Seoul, South Korea, 2008
S. X. Ding, P. Zhang, C. I. Chihaia, W. Li, Y. Wang and E. L. Ding
- Fault detection of networked control systems with packet based periodic communication. International Journal of Adaptive Control and Signal Processing, 2008
Y. Wang, S. X. Ding, H. Ye, L. Wei, P. Zhang und G. Wang
- Fault detection of networked control systems with packet loss. IFAC World Congress, Seoul, South Korea, 2008
Y. Wang, S. X. Ding, P. Zhang, W. Li, H. Ye and G.Z. Wang
- Integrated design of an observer-based fault detection system over unreliable digital channels. 47th IEEE Conference on Decision and Control, Cancun, Mexico, 2008
W. Li and S. X. Ding
- Networked Fault Detection Systems with Noisy Data Transmission. Automatisierungstechnik 56 (1), 2008
W. Li, P. Zhang, S. X. Ding, C. I. Chihaia, E. Goldschmidt, O. Bredtmann und A. Czylwik
- A hands-on introduction to Real-Time Linux AKS internal report, Mai, 2009
Cristian I. Chihaia
- Decentralized fault detection of large-scale complex systems. IFAC SAFEPROCESS, Barcelona, Spain, 2009
W. Li, W.H. Gui, Y.F. Xie and S.X. Ding
- Fault detection of networked control systems with limited communication. International Journal of Control, März 2009
Y. Wang, H. Ye, S. X. Ding, Y. Cheng, P. Zhang und G. Wang
- Observer-based fault detection of technical systems over networks Mai, 2009
Wei Li
- On fault detection of networked control systems. IFAC SAFEPROCESS, Barcelona, Spain, 2009
W. Li and S.X. Ding
- Residual generation an evaluation of networked control systems subject to random packet dropout. Automatica, 45, 2427 - 2434, 2009
Y. Wang, H. Ye, S. X. Ding, G. Wang, D. Zhou
- WiNC - Advanced Experimentation Platform for Wireless Networked Control. IFAC SAFEPROCESS, Barcelona, Spain, 2009
C. I. Chihaia, O. Bredtmann, E. Goldschmidt, W. Li, S. X. Ding and A. Czylwik
