The Research On Fault Diagnosis And Maintenance Optimization Of Rotary Vane Pump Based On SVM

In mechanical manufacturing of products,pump industry is becoming one of the most important parts of it,and more and more people are focused on pumps' performance,stability and efficiency of maintenance.Following the constant improvement of technology of fault diagnosis,because of restriction of limited fault samples,complexity of algorithm and habit of human element maintenance,artificial intelligence fault diagnosis is difficult to applied in enterprises.Owing to superiority of algorithm and characteristic for the method of support vector machine,the paper establishes a fault diagnosis model for rotary vane pumps.The model is not only overcomes the low classification accuracy due to the limited fault samples,but also improves efficiency of maintenance of rotary vane pumps.In the face of the problem about taking long time for fault diagnosis in the process of maintenance of rotary vane pumps,it is becoming imperative to establish intelligence fault diagnosis model and implement application for rotary vane pumps.Aim at the above problems,the paper proposes global optimization classification of support vector machine method for fault diagnosis model and takes full analysis on the effect of fault classification accuracy and running speed based on fault characters of rotary vane pumps and experience of human element maintenance.Then the paper further proposes an improved SVM global optimization classification algorithm based on conjugate transformation for the model of fault diagnosis of rotary vane pumps to further improves classification accuracy and efficiency of maintenance for rotary vane pumps:Firstly,the paper seeks for the comfortable support vector machine algorithm for the fault diagnosis of rotary vane pumps based on the analysis about characteristics of each SVM algorithm and pumps,then classifies and trains the maintenance fault samples,analyzes classification accuracy and running speed in different selection of kernel function and data format,ultimately gets the optimal SVM fault diagnosis model for rotary vane pumps.Secondly,by acquainting with present maintenance conditions about the enterprise,the paper uses queue theory to analyze the reseaonability for maintenance conditions.Through comparison with time of optimal fault diagnosis model on SVM method,ultimately gets the promotion time for maintenance,so as to optimize the maintenance time.Finally,the paper proposes the concept of key fault characteristic units based on fault characters of rotary vane pumps.On this basis,an improved SVM global optimization classification algorithm based on conjugate transformation is proposed to optimize the input for support vector machine,ultimately further improve efficiency of maintenance and optimize the process of maintenance.