Research On Fault Diagnosis Of Wireless Sensor Networks Containing Disturbances

Mobile Wireless Sensor Networks(MWSN)is one of the most influential key technologies in today's society.It has been widely used in military surveillance,medical care,assisted living,and environmental science.MWSN is constructed by a large number of movable nodes through self-organization,and integrates wireless communication technology,sensor technology,and distributed information processing technologies.It has a large number of nodes,wide distribution range,limited power supply energy,and complex application environment.characteristic.Under the influence of these factors,compared to the fixed wireless sensor network,the nodes in MWSN are more prone to failure.The failure of the node will affect the connectivity of the entire network,and the failure of a large number of nodes will cause the network to be paralyzed.Therefore,it is of great significance to study the fault diagnosis of mobile wireless sensor networks.This paper studies the fault diagnosis method of mobile wireless sensor network,the main work includes the following aspects:1.This article first analyzes the node structure,network structure and fault sources in MWSN,and divides the node faults in MWSN into power module faults,communication module faults and sensor module faults,and transforms them into mathematical models.The corresponding network model,energy consumption model and mobile model are established for the types of common sensor nodes and sink nodes in the network.On this basis,NS3 software is used to simulate the MWSN to collect data,and the fault is added to the collected data.2.A MWSN fault detection method based on random forest is proposed.The method is based on the random forest(RF)algorithm.Its voting method is improved to reduce the influence of decision trees with poor classification ability on the output of the entire model.And through the CSA algorithm to optimize the parameters.Divide the proportion of different failure rates and conduct experiments.The results show that compared with the original random forest algorithm,this method has a higher accuracy rate in MWSN failure detection.3.A MWSN fault classification method based on deep forest is proposed.This method is based on the deep forest(DF)algorithm.The cascade forest part is improved to reduce the input vector dimension when the number of cascade layers is high.And make full use of the classification results of each layer of forest,combined with fault detection methods to complete fault diagnosis.Divide the proportions of different training sets and conduct experiments.The results show that,compared to directly classifying the fault data,this method can more accurately identify the fault in the MWSN fault diagnosis.