Wireless Sensor Network Node Module Level Fault Diagnosis Method Research And Application

Based on the rapid development of microelectronic technology, communicationtechnology, embedded technology, sensor technology and power technology, theapplication scope of wireless sensor network (WSN) technology is greatly expanded. Inorder to improve the reliability of wireless sensor network system and the easy diagnosticof node fault, an airport fuel insurance system and a wireless security system of chemicallaboratory are chosen as research objects in this dissertation, the research of module-levelfault diagnosis method for WSN nodes is carried out, a new fault diagnosis method isproposed, and it has been applied in different physical locations. Therefore, the scientificresearch in this dissertation has important theoretical significance and practical application.For muggy, vibration, electromagnetic interference, inflammable, explosive, and toxicenvironment, in order to improve the reliability of WSN nodes work and the accuracy ofdata aware, in this paper, the actual operating conditions of wireless monitoring system formobile refueling device in certain airport and wireless security monitoring system inchemical laboratory are analyzed, an online-status monitor in WSN nodes is designed, andthe fault diagnosis methods based on Gaussian process regression, fuzzy neuralnetwork(FNN) are proposed. The main research work in this dissertation is summarized asfollows:1. In order to quickly detect the fault locations in WSN nodes and diagnose the causeof faults, an online-status monitor of WSN nodes modules fault is designed, and it willrespectively online monitor the status of sensor module, processor module, wirelesscommunication module and power supply module in WSN nodes. Each module fault typeis divided. The fault signature and the fault location matrix are introduced. And the faultreason diagnosis table is established. The fault isolation method is applied to the design ofWSN nodes, and the failed node module is effectively isolated in order to prevent the rangeexpansion of nodes failure.2. Aiming at easily failed wireless communication module, the current characteristicmodel of wireless communication module in different states is established. The faultdiagnosis parameters of wireless communication module in different states based onGaussian process regression are dynamically adjusted. Experimental results show that thefault diagnosis method can not only diagnose short circuit, open circuit, current overdrawn, too small current and other faults in wireless communication module, but can also diagnosedifferent fault types such as transmit power reduced or zero power output caused byamplifier circuit aging, burning or other causes in wireless communication module.3. A fault diagnosis method of wireless communication module based on fuzzy neuralnetwork (FNN) is proposed. FNN parameters through off-line training is obtained,combined with AD convert precision, wireless communication module current waveformanalysis and filtering method, and the fault diagnosis model is applied to the actualoperating WSN nodes. Experimental results show that the calculation of fault diagnosisbased on FNN model is smaller and the accuracy of fault diagnosis is higher.4. In order to quickly and accurately diagnose WSN nodes faults, an easy diagnosisstrategy of WSN nodes faults is proposed and a WSN node of fault easy diagnosis isdesigned. Installing nodes status monitor, obtaining fault information and replacing faultmodule become convenient. Meanwhile, in order to reduce the effects of noises on WSNnode status monitor, a composite filter combining wavelet de-noising and median filteringwas used. This method is not only able to filter out most of the interference, making thewaveform becomes smoother, but also retain more signal edge. In order to extend the lifecycle of WSN nodes and effectively save the nodes energy, experimental tests for receiving,transmitting, monitoring, sleep current consumption and time characteristics in wirelesscommunication module are carried out. The nodes sending, receiving interval, every timesending, receiving byte count are introduced to further subdivide the network loadparameters. The status probability of node receiving a signal is analyzed and thecommunication energy consumption model is established. By taking this, under differentnetwork load and receiving interval, the optimal energy consumption parameters can beachieved. Meanwhile, in the subcontract transmission of large amount data, the first packetof data is added with an appointment frame, so that the receiving node falls into sleep afterreceiving a plurality of data packets consisting of complete files. Therefore, it avoided theunnecessary multiple rouse and obviously reduced the transmission energy consumption.5. The wireless monitoring system of airport mobile refueling devices and the wirelesssecurity monitoring system of chemical laboratory are designed. The fault diagnosismethods proposed in this dissertation are respectively applied to these two systems. Theexperimental results show that the fuel trucks will upload transport condition, refuelingschedule, ground wells and other information to the dispatching centre for review, confirmand schedule plan process. Using fault diagnosis methods can accurately diagnose the location of nodes failure and determine the fault module of internal nodes and failurecauses. In the process of WSN nodes online monitoring toxic gas content, concentration ofsmoke particles, flames and other information, the fault diagnosis methods can accuratelydiagnose the position of fault nodes installation, determine the failure module of theinternal nodes and fault causes. As fault diagnosis methods applied in the two systems,multiple types of failures in WSN nodes are detected and the reliability and practicability ofWSN system is improved.