Research And Application Of Multi-Attribute Fault Data Classification Method

The basis of safety management and safety decision-making is on-the-spot analysis of breakdown when the system is operating.No matter what kind of security management domain it is,the security management objects can all be abstracted as a system structure.Then the basis of security management and decision-making is the basic data,system abstract method and data analysis method.Because the collection of fault data is affected by human factors,system defects and measurement errors,the data are discrete,fuzzy and random,no matter from the experience of field monitoring equipment or operators,in particular,empirical data on personnel.Field operators can provide the most direct experience data,which reflects the system security characteristics and personnel experience,and is also a valuable system security analysis materials.Therefore,a method is needed to analyze the system fault information provided by field operators.However,due to the differences in operator experience,system operating environment and other reasons,the current method is still difficult to effectively analyze.Based on the idea,theory and method of spatial fault tree,factor space and multi-attribute decision-making,this paper presents a multi-attribute breakdown data classification method based on the coupling of Monte Carlo method and attribute circle.The main findings and conclusions are as follows:(1)This paper presents a multi attribute fault data classification method based on the coupling of Monte Carlo Method and attribute circle.By using Monte Carlo method simulation,the area of attribute circle,object attribute and object overlapping area is calculated.By using the idea of distance between two points in multidimensional space and object overlapping degree,calculation of object similarity and generation of similarity matrix,object similarity analysis are carried out.(2)Classification of system fault data is completed based on the algorithm model.The characteristics of actual safety management data are discussed,and the system safety is measured in breakdowns,and the Algorithm model is analyzed with 30 classified records of system operating safety faults as the research object.All the classification objects of the Algorithm meet the requirements of direct correctness and adjacent correctness,so the algorithm has a high matching degree.The results prove that the classification method and model proposed in this paper can improve the accuracy and efficiency of fault data classification,reducing failure rate and improve management quality.(3)This paper draws attribute range distribution graph and calculate the barycenter value of attribute range with 1-d barycenter method.It analyzes the attribute range distribution of the environment conditions corresponding to each kind of security,and draws the attribute range distribution graph according to the different attributes of the classified objects,combined with one-dimensional barycenter method.Also,the maximum barycenter and minimum barycenter of each attribute range of each object in different security classification are calculated.Furthermore,it can be determined that the set of object taxonomies composed of these objects with similar security is suitable for the conditions of the working environment,then in the corresponding environment,the objects in these object taxonomies prove to work safely and reliably.This paper belongs to the field of combining theory with practice and combining management science with safety science,and develops the spatial fault tree theory and factor space theory.The research can provide useful methods for the development and application of information science and data science,as well as new basic theories and methods for fault experience mining and extraction in multi-attribute data.The dissertation has 19 figures,13 tables and 131 references.