Research On Text Multi-tag Classification Model Of Rail Transit Equipment Failure

At present,the maintenance and repair of equipment faults are generally based on the experience of the workers in combination with their own maintenance experience to develop a variety of maintenance measures,and maintenance workers record data for various fault conditions are mostly based on personal experience dictation records,most of the data In the form of short texts,it is difficult to standardize management and quantitative analysis of equipment fault data conditions because it is impossible to effectively use equipment fault information.This article will systematically analyze the textual data of equipment failures in a subway company,apply the text analysis technology to the field of rail transit fault data analysis,and improve the current waste data text resource waste situation by relying on scientific theories,starting from the equipment failure data.Construct a fault classification model that meets the characteristics of maintenance data,and conduct standardized management of fault data.The text constructs a keyword word bank of subway equipment faults by integrating the subway fault data based on the map ranking method.At the same time,a word vector representation model based on Bi-LSTM is constructed based on the preliminary expression of word vectors in word2vec.The word vector of the feature keyword of the training sample is represented,and the word vector can more effectively reflect the contextual feature word relationship.Finally,in order to reduce the amount of training for the classifiers,the DAG-SVM multi-label classification model was improved.The two categories of the original multi?category were compared and classified to improve the multi-round elimination training method.Each round of training only required adjacent The two categories can be classified and voted out,thereby eliminating the one with the least votes in the current round of the word,so that the number of classifiers can be improved from n to n,reducing the amount of calculation.For the quantitative analysis of the next fault,the rational use of statistical data to develop adjustment and maintenance plan to provide the data basis.The research in this paper will solve the problem that the current failure data of a subway equipment is difficult to use effectively,and it cannot provide an effective scientific theoretical guarantee for the equipment maintenance plan.The fault-type keyword dictionary proposed in this paper will improve the contrast analysis standard of subway equipment failure and provide a standardized strategy for equipment failure data.The text also proposes a Bi-LSTM-based word vector representation model,which extracts the key features of the fault text data and converts it into a trainable vector representation.This model compares the accuracy of existing traditional word vector algorithms in the training of faulty text data.Can be about 10%,more in line with the needs of equipment failure text data features.On this basis,an improved DAG-SVM multi-label text classification model is proposed.Provides an automated classification analysis scheme for equipment failure data.Provides a systematic data foundation for future equipment failure data processing,and improves the utilization of fault data.Finally,the experiment verifies that the classifier proposed in this paper is superior to the traditional classifier in the accuracy of the premise,the calculation speed will be more excellent with the scale of the class label increases.