Redundant Overall Structure And Optimization Methods For Robust Sensor Network In Nuclear Power Plant Environmental Stressors

Energy crisis and environment pollution are worldwide issues, and it is significantto develop abundant and clear energy, such as nuclear, wind, and solar energy, amongwhich security issue of nuclear energy is extremely vital. Obviously safe and reliableoperation of nuclear power plant (NPP) is important problem. Current wired monitoringtechnology has some deficiencies like wiring constraints, high cost, and maintenancedifficulty. Applying wireless technology to NPP can solve these deficiencies for itsflexibility, low const, and maintenance convenience. However, there are twocontradictions between wireless application and NPP environmental stressors. On theone hand, bad NPP environment such as electromagnetic interference, wirelessinterference, and multi-path effect can reduce the quality of wireless communication; onthe other hand, wireless signal may affect sensitive NPP devices. To address thecontradictions, current research mainly focus on two methods, one is weak signalanti-interference modulation technique on wireless sensor network(WSN) physical layer,and the other is to improve wireless network performance through optimization designon WSN network protocol above physical layer. There are some researches about thesecond method, but the solution is not systematic and not applicable for NPP. Thisdissertation presents related systematic research, and proposes a series of strategies andalgorithms is to improve network performance.Current research focuses on wireless node placement, wireless link scheduling,routing protocol, and optimal search etc, and these are not applicable for NPP deviceshealth monitoring application. Thus this dissertation focuses on the above research aswell as NPP application, and proposes some novel algorithm on datalink layer, networklayer, and application layer. Experimental results show that these algorithms can helpestablishing robust and real-time WSN with long network life time in NPP applicationwith bad environmental stressors.A novel WSN relay nodes placement strategy in NPP application is proposed. Itconsiders NPP special problems, such as obstacles blocking, low network robustness,directional antenna technology, and constrained placement positions. This problem is formulated into an NP hard problem, and a three-step solution, which is establishingredundant edge-connected network, genetic algorithm optimization, and generatingrelay nodes positions respectively, is used to solve this problem. Experimentationdemonstrates that these strategies outperforms practically used placement strategieswith3-D obstacles avoidance, better fault tolerance, and optimized placement position,which can solve relay nodes placement problems in NPP application.Vibration devices modal analysis and genetic algorithm is used to solve vibrationmonitoring nodes placement problem. Through analysis on current WSN monitoringnodes placement research and corresponding NPP application requirements, forexample modal information should be maintained, the monitoring nodes should notinfluence sensitive NPP devices, the monitoring nodes should be connected, and theplacement positions should be secure, easy for deployment and for maintenance.Through finite-element analysis, modal information is obtained. Considering otherconstraints, this problem is formulated into a combinational optimization model. Andthen, suitable genetic algorithm operators, parameters, and fitness function areconstructed. Comparison with current practical placement strategies indicates theeffectiveness of our strategy.This dissertation proposes two heuristic communication resource schedulingalgorithms for low energy consumption and low cost WSN. The first algorithmformulates the scheduling problem into a graph vertex coloring problem, and proposedan improved Welsh-Powell algorithm to obtain approximate solution. The secondalgorithm is fast and easy to implement in application, and also considers more practicalissues. Both algorithms solve communication interference, flow balance, and paralleledresources use. These two algorithms are compared with typical greedy algorithm andhybrid intelligent algorithm in terms of superframe length, running time, throughput,delay; and experimental results proves that they achieve balance between solutionquality and efficiency.A multi-link scheduling strategy using power control algorithm is proposed forhigh performance WSN. By referring Foschini’s research on power control andconducting proper relaxation to this problem, a heuristic algorithm called BSP isproposed to solve this first stage problem; and then for unscheduled links, an efficienttwo-stage assignment algorithm is proposed to reschedule these links through binary search. This algorithms-ECT achieve high performance in comparison withapproximation algorithm. In addition, experiments also shows that second stagescompensate quality loss at the first phase, the efficiency of algorithms have improvedgreatly.With bad NPP environmental stressors, the data transmission may be unreliable,and some NPP instrument is sensitive to radio frequency signal. To meet reliabletransmission and long network life time requirements, a location based multi-pathrouting algorithm (LBMR) is proposed. This algorithm takes advantage of the relativelyfixed positions of nodes in NPP application and refers other multi-path and hybridrouting protocol. The routing protocol based on LBMR constructs routing cost functionby an integration of link quality, link payload, and sensitive NPP instrument. Thisprotocol establishes routing searching zone and uses Dijkstra algorithm to calculatedisjoint redundant paths. Also it improves packet transmission process and packetbuffering strategy. In comparison with single-path location-based routing and othermulti-path routing protocols, our routing protocol achieves high performance onnetwork robustness, real-time, and network lifetime.In wireless sensor network, there are many flooding process during timesynchronization, network formation, routing search, data transmission, sensor queryingetc. The flooding process is easy, but it is often low-efficient and unreliable. A localoptimal search algorithm based on optimal search theory is proposed. This algorithmfocuses on cluster network topology, and store cluster label in sink node. The searchingfunction is optimized; the cluster label sequence and cost distribution function isgenerated through searching theory. Experimental result shows that our searchingalgorithm achieves better performance than flooding and current local optimal searchingalgorithm in large-scale WSN application. In addition, the searching resource allocationproblem of data active collection process is transformed into a mobile agent searchingresource allocation problem, and a global optimal resource allocation model is proposed.Experiment result indicates that this algorithm achieves better resource utilization incomparison with commonly used resource allocation algorithm.