Design And Implementation Of Electronic Product Fault Maintenance Assistant System

In daily life,because of the frequent occurrence of common electronic product failures,it is an important link in after-sales service to determine the causes of electronic product failures.At present,the maintenance of electronic product fault is generally based on the maintenance measures formulated by the maintenance workers according to the fault conditions and their own maintenance experience.Most of the fault status data recorded by the maintenance workers are from the user's oral statement and their own experience.Most of the data are in the form of short text,so it is difficult to carry out standardized management and quantitative analysis of the equipment fault status.Therefore,the text data of equipment related faults in a after-sales service system is systematically analyzed,and the short text analysis technology is applied to the field of electronic product fault data analysis.Starting with the fault data of equipment,the fault classification model that conforms to the characteristics of maintenance data is constructed,and the auxiliary system of fault maintenance is designed and implemented to support the standardized management of electronic product fault data,it has good practical significance and application value.In this thesis,the short text of fault description and its label in an electronic product aftersales service system are used as data sources for short text classification.Short text vector model is constructed by short text preprocessing and Feature Engineering,the short text classification algorithm is designed and realized,and the experimental results are analyzed by combining the classifier evaluation index,the hierarchical model is constructed and the electronic product fault maintenance assistance system is designed and realized.The main contents of this thesis are as follows:(1)The short text information and category information of electronic product fault in a database of after-sales service system are obtained to build a corpus of electronic product fault,and the short text preprocessing is carried out to provide the basis for the design and implementation of electronic product fault prediction auxiliary system.(2)After short text preprocessing,words need to be represented and feature selected.In the hierarchical model and non hierarchical model,ML-KNN,weighted SVM,LSTM and Text CNN classifiers are used respectively to compare TF-IDF and Word2 Vec methods.It is found that Word2 Vec method has better classification effect in the electronic product fault short text classification model.(3)Aiming at the hierarchical classification characteristics of electronic product fault data,this thesis applies the hierarchical model to the problem of electronic product fault short text classification.Through the experiment comparison,we propose the electronic product fault stratification model which uses LSTM classifier in the upper level and Text CNN classifier in the lower level.(4)Based on the above work,this thesis designs and implements an electronic product fault maintenance assistant system.The system uses Word2 Vec method for fault short text representation and feature selection,and uses LSTM + Text CNN combination to build fault short text hierarchical model.The system test results show that the designed and implemented fault maintenance auxiliary system meets the actual needs.