Research On Fault Diagnosis Method Of Wireless Sensor Network Node In Interference Environment

Wireless Sensor Network(WSN)have the advantages of low power consumption and wide coverage,making them widely used in the monitoring of industrial production environments,which is of great significance for ensuring industrial safety.WSN nodes are typically deployed in complex working environments,and sensor node failures are inevitable.In order to timely detect sensor node failures,ensure the reliability of data collection,and guarantee the safety of industrial production,fault diagnosis of WSN nodes plays an important role in safeguarding industrial production safety.Existing WSN node fault diagnosis methods often require extracting node feature data as the basis for fault diagnosis,and the diagnosis methods can be broadly divided into three categories:model-based analysis methods,data-driven methods,and hybrid information-based methods.Belief Rule Base(BRB)is a type of hybrid information-based method that has the advantages of small sample training and strong interpretability,which are in line with the characteristics of WSN node fault diagnosis with low sample data and high reliability requirements.It is a very effective method for WSN node fault diagnosis.The feature data of WSN nodes are affected by the complex working environment and are therefore unreliable,which has a great impact on the accuracy of node fault diagnosis.Secondly,as the node’s working state changes,the discriminability of feature data to fault types is uncertain and can be high or low,and existing fault diagnosis methods often place more emphasis on changes in a certain feature data.When the discriminability of that data to fault types decreases,the accuracy of fault diagnosis is reduced.In addition,there is similarity in feature data between some faults,which can easily lead to local unknown fault types,and the diagnosis model cannot effectively represent such fuzzy information,which also affects the effectiveness of fault diagnosis.In order to improve the accuracy of fault diagnosis for WSN nodes,this paper conducts research on fault diagnosis methods for WSN nodes under interference environments,with the Belief Rule Base(BRB)as the theoretical basis.The paper makes innovative contributions in three aspects: handling the unreliability of feature data,considering the uncertainty of feature data in distinguishing fault types,and dealing with local unknown fault types.(1)To reduce the negative impact of unreliable feature data in a noisy environment on the fault diagnosis of wireless sensor network(WSN)nodes,a fault diagnosis model called Belief Rule Base with Self-Adaptive Quality Factor(BRB-SAQF)is proposed based on the BRB method.The model innovatively introduces the attribute quality factor to eliminate the effect of unreliable feature data on WSN node fault diagnosis,effectively reducing the negative impact of unreliable feature data.(2)To address the uncertainty of the distinguishing ability of feature data for fault types in WSN nodes,an adaptive attribute weight Belief Rule Base(BRB)method is proposed for fault diagnosis,called BRB with Adaptive Attribute Weights(BRB-AAW).Unlike traditional BRB methods,which use a single set of attribute weights for all rules,this model independently sets attribute weights for each rule,which reflects the true distinguishing ability of feature data for fault types in different WSN operational states,and improves the accuracy of the WSN fault diagnosis model.(3)The paper proposes a fault diagnosis model for wireless sensor network(WSN)nodes based on the extension of the traditional BRB fault identification framework to the power set form.This is done to effectively handle local unknown fault types caused by the similarity of feature data.By mapping local unknown fault types to subsets of the identification framework,the model can more effectively represent these types and improve the accuracy of WSN node fault diagnosis.This results in the creation of a fault diagnosis model based on a power set belief rule base(PBRB).(4)To facilitate the quick usage of the three proposed models for WSN node fault diagnosis,a prototype interference environment WSN node fault diagnosis system was designed and developed.The BRB-SAQF module is used to calculate the reliability of sensor feature data,the BRB-AAW module is used to construct a rule base with independent attribute weights,and the PBRB module is used to improve the model’s ability to represent local unknown fault types.Through the research on fault diagnosis methods for WSN nodes in interference environments,three fault diagnosis models were proposed based on the BRB modeling method,which addressed three aspects: handling the unreliability of feature data,considering the uncertainty of feature data in distinguishing fault types,and handling local unknown fault types.The innovative introduction of attribute quality factor parameters effectively reduced the negative impact of unreliable feature data on WSN node fault diagnosis in interference environments,resulting in better convergence of diagnosis results.The method of independently setting attribute weights for each rule more accurately reflected changes in the distinguishability of feature data for fault types,improving the diagnosis accuracy of the model.The extended power set identification framework more reasonably represented local unknown fault types,improving the modeling and fault diagnosis capabilities of the model.