Wireless Sensor Network based Data Fusion and Control Model for an Oil Production Platform

Wireless Sensor Network based Data Fusion and Control Model for an Oil Production Platform

Mitun Bhattacharyya (R. V. College of Engineering, India), Ashok Kumar (University of Louisiana at Lafayette, USA) and Magdy Bayoumi (University of Louisiana at Lafayette, USA)
DOI: 10.4018/978-1-61520-843-2.ch006
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

In this chapter the authors propose methodologies for improving the efficiency of a control system in an industrial environment, specifically an oil production platform. They propose a data fusion model that consists of four steps – preprocessing, classification and association, data association and correlation association, and composite decision. The first two steps are executed at the sensor network level and the last two steps are done at the network manager or controller level. Their second proposal is a distributed hierarchical control system and network management system. Here the central idea is that the network manager and controller coordinate in order to make delays in feedback loops as well as for increasing the lifetime of the sensor network. The authors finally conclude the control system proposal by giving a controlling model using sensor networks to control the flow of hydrocarbons in an oil production platform.
Chapter Preview
Top

There are a few related works in this field from the control system point of view. These works define the sensor network as a component of the system with varying delays. The work in (Eriksson & Koivo, 2005) gives a model for tuning of a PID controller assuming a Gaussian varying time delays. The work in (Sinopoli, Schenato, Franceschetti, Poolla, & Sastry, 2005) analyzes a control system with time varying delays between sensor and controller and controller and actuator. The work in (Colandairaj, Irwin, Scanlon, & William, 2007) explores a combination of data rate scaling and sample rate adaptation for control systems to handle time varying network conditions. The other works include (Kawka & Alleyne, 2005), (Adam, Brady, & Kosc, 2001), (Peng, Huijin, Lei, Zhi, & Anke, 2006). In (Nikolakopoulos, Panousopoulou, Tzes, & Lygeros, 2005) a LPQ controller is designed whose parameters are changed depending on number of hops taken in network. Having a system based on complete path knowledge from source to destination is expensive and not implementable in sensor networks. Moreover, the Zigbee and 802.15.4 standards are used in sensor networks. According to (Nixon, Deji, Blevins, & Mok, 2008), control loop should execute 4 to 10 times faster than the processing time plus dead time. In addition to delays in the control loop, there is also a presence of jitter (i.e., variation in inputs, outputs, control execution). In terms of communication, the variation in delay to deliver a measurement from source to destination is referred to as jitter. By synchronizing network related processes with control execution, the need for over-sampling and communicating excessive measurements can be minimized. The work in (Dei, Snickars, Landernas, & Isaksson, 2008) proposes a methodology to counteract controller clock drift. It is seen that the performance improves if a Predictive PI (PPI) controller is used. The work also discusses a simulator tool in Simulink called TrueTime. The work in (Lennvall, Svensson, & Hekland, 2008) gives more details of the TrueTime Simulator.

Complete Chapter List

Search this Book:
Reset