Multi agent system to monitor structures
Download as PPTX, PDF1 like680 views
1) The document proposes a multi-agent system called NMAOS to develop an agile wireless sensor network for structural health monitoring. 2) NMAOS is a self-learning system that can predict peak loads on a structure and trigger messages to activate the sensor network during unexpected incidents. 3) The multi-agent architecture includes triggering, comparative, sensor, collaborating, damage diagnostic, and prognostic agents to monitor the structure and assess damage.
Multi agent system to monitor structures
- 1. Multi Agent System for Agile Wireless Sensor Network to Monitor Structures Othman Sidek and S. A. Quadri CEDEC (Collaborative Microelectronic Design Excellence Centre) Universiti Sains Malaysia, Engineering Campus 14300, Malaysia Paper presented at PIERS Conference , Proceedings Marrakesh, Morocco, Mar. 20-23, 2011 765
- 2. Keywords : Multi agent systems Structural health Monitoring Wireless sensor networks Agility
- 3. •Multi-agent system (MAS) is a system composed of multiple interacting intelligent agents. •An intelligent agent (IA) is an autonomous entity which observes and acts upon an environment. •Intelligent agent learns and use knowledge to achieve goals.
- 4. Wireless Sensor Network Wireless Sensor Network (WSN) consists of spatially distributed autonomous sensors to monitor physical or environmental conditions.
- 5. Agility Agility is defined as the capability of a WSN to timely capture the event response of the monitored structure, process the measured data, extract the relevant features, and interpret the results.
- 6. Why need an agile WSN ?
- 7. Golden Gate Bridge (GGB), California University of California, Berkeley, installed 256 accelerometers to monitor bridge.
- 8. The sensor network on the bridge did not collect any data during earthquakes in 2006 because the monitoring system was in Passive mode !!! The significant non-linear behavior analysis at extreme-conditions could not be studied because of non availability of data.
- 9. Thus in order to ensure reliable WSN system to monitor structure a multi agent model is proposed.
- 10. NMAOS N => Never M => Miss A => An O => Opportunity S => Scheme
- 11. Practical back ground / Previous work
- 12. Non-preemptive MAC without priority support has failed to capture signals during earthquake.
- 13. Liang Cheng and Shamim N. Pakzad ( IEEE Conf, 2009) proposed a Pulse-based media access control scheme. A trigger message from a nearby observation site can be timely propagated across a WSN to preempt current tasks such as energy-saving sleeping and scheduled data transmissions so that the sensor network can be forced into a record ready state before the high impact load (earthquake) waves reach the monitored bridge.
- 14. NMAOS (never miss an opportunity scheme) It is self-learning intelligent-based software system which obtains the training data by observing the real time traffic on a particular site (bridge), and by efficiently analyzing the captured data the system predicts maximum load on the structure at a particular instance of time, thus can trigger messages promptly to make the system active. Thus ensures the WSN system to be active during unpredicted incidents.
- 15. To apply multi-agent technology to a distributed SHM system, each component or subsystem in the SHM should is changed to an agent. Six kinds of agents are defined to develop an agile WSN for a SHM system.
- 16. The multi-agent architecture for an Agile WSN
- 17. Triggering agent: It triggers the system by generating the pulse A self-learning intelligent-based software system predicts the maximum peak load on the structure (bridge) at some particular time, and trigger more priority pulses to active monitoring system.
- 18. Comparative agent Comparative logic is embedded in the comparative agent, which accomplishes the task of preempting current tasks, such as energy-saving sleep and allows trigger pulses to activate the whole system to monitor the structure during any disastrous event.
- 19. Sensor agent The sensor agent is responsible for sensing parameters, such as stress, strain, pressure displacement, acoustics, and temperature. They accomplish many tasks such as data processing and disseminating.
- 20. Automatic collaborating agent Supports low-power, multi-point, and heterogeneous operations with a distributed synchronization mechanism. It takes care of clustering, fusing and communicating data exchange between different agent entities.
- 21. Damage diagnostic agent It comprises a real-time automated reasoning and decision-making integrated software system. Training data is fed to Artificial Neural network to incorporate Expert decision-making.
- 22. Prognostic agent Prognostic agent evaluates the remaining life- time of the structure at a given state of damage and future loading. The future load spectra can be either damage tolerance or safe-life philosophy. The results of the damage diagnostic agent provide information on the current state of the structure for prognostics.
- 23. CONCLUSION AND FUTURE WORK Agility of a WSN is an important factor contributing to reliable and successful SHM programs. The prototype NMAOS (never miss an opportunity scheme) based on multi-agent technology proposed uses the PB-MAC based technique of triggering a message of high priority to preempt the current passive state of the WSN and make it active for sensing and capturing data at the time of an unpredicted event of high impact on the structure, such as an earthquake. Further studies can focus on evaluating the effectiveness and validating this cognitive multi-agent based prototype in real-time situations with a view to developing an agile wireless network system to monitor the health of structures.