The Research On Fault Detection Methods In Wireless Sensor Network

Wireless sensor network is the main means of the objective world perception, state detection and distributed management, has become the supporting technology of the Internet of things, perception of the earth. But the uncertainty of sensor networks, such as sensor failure, communication failure and node failure, limit the extensive application of sensor networks. Especially the fault nodes of the error data, if it is not treated and used directly, will cause unpredictable damage. Therefore, it is necessary according to the node and the whole network to study a set of fault detection, diagnosis, health evaluation method, the node can be their own health status and sensitive data sent to the base station base station together, according to the health level of the whole network, establish fault tolerant control strategy, to maximize the application of cyber source, maintenance of network the long-term stable operation and prolong the lifetime of network etc.,This paper mainly includes two aspects, respectively:1) Self-detection method based on node fault online data. Time series analysis technique based on the operation of the components, data acquisition, data sensor nodes as input, and the preprocessing of data (the elimination? trend, zero mean processing), and then combined with the time series analysis method, node data prediction. The actual operation data by comparing the model predicted data and the next cycle, judge the nodes of the current level of health.2) Node fault diagnosis method based on TTDPN model. The research based on the TTDPN (T-Timed Discrete Petri Nets) of the sensor network node modeling method, a sensor network node working process of TTDPN model, the node fault diagnosis method based on TTDPN model, the functional positioning and diagnosis of sensor node failure, and further use of TDDPN model for fault recovery and health management and evaluation.Through the study of this subject, to faults in the sensor nodes and the entire network formation of self-detection strategy, improve data reliability and robustness. According to the sensor nodes within the data, real-time fault detection, the closed-loop control node state of health, the effective filtration pathological data, only to send sensitive data is correct and effective. To improve use of cyber source rate, improve sensitive data, extended sensor network applications. Study on fault diagnosis of sensor network recovery and method of health management, self-recovery capability of sensor nodes and the whole network survivability, maximizing the network deployment and application effect cost ratio.