Research On Fault Diagnosis Method Of Wireless Sensor Network Based On Extreme Random Tree

Wireless Sensor Networks(WSN)is one of the key influential technologies in today's society and are widely used in areas such as military surveillance,healthcare,assisted living and environmental science.WSN act as an interface between the physical and digital worlds and can lead to serious consequences in terms of safety,economy and system reliability when sensors pass incorrect data to aggregated nodes.Therefore,it is important to study the problem of fault diagnosis in wireless sensor networks.In this context,this paper investigates fault diagnosis methods for wireless sensor networks,and the main work includes the following aspects:Firstly,the basic structure,network model and node structure of WSN are introduced.By analyzing fault sources,common faults of WSN nodes are classified,and feature maps and mathematical models of different faults are presented.A data acquisition system based on Ali Cloud platform is designed to collect data,and various faults are injected into data according to the proposed mathematical model,providing data support for subsequent algorithm research.Secondly,a WSN fault diagnosis method based on extreme random tree is proposed,which is based on extreme random tree algorithm.Its decision tree training method is improved to enhance randomness and reduce the influence on result prediction.The weighted voting strategy is formed by setting different weights for the decision trees with different classification accuracy.The sensitive parameters of the algorithm were selected by cross validation and simulated annealing algorithm was used for parameter tuning.The feasibility of the improved algorithm was verified by injection fault data set experiment.Finally,the data set collected by the system after fault processing is used to conduct experiments on the simulation platform to verify the improved effect of the proposed algorithm,and it is compared with four machine learning algorithms for WSN fault diagnosis,such as support vector machine,neural network,random forest and decision tree.Three evaluation indexes of accuracy,accuracy and F1 score were selected to compare the performance of the algorithms in the simulation experiment.The results show that the proposed method has high performance in WSN fault diagnosis.