Research On Degeneration Assessment & Trend Prediction About Machine Based On Logistic Regression And SVM

Development of Information Technology drives industries towards continuous, high-speed and large-load. On one hand, the development lowers product cost, improves efficiency, speed up product design, manufacturing and service. On the other hand, the development needs higher level design, manufacturing, and the ability of work and maintenance. Furthermore, each enterprise has to cut down the fault time and lengthen the effective life cycle to strengthen their marketing competition. So we proposed the system, which can assess the machine running condition and predict the changing trends. So we can maintain and repair the machine or equipment. Therefore, the fault time will be shortened and the effective life lengthened. The paper composed of the following parts:(1)Such machines statuses as normal, different mass unbalance, radial rub are simulated for validating the effectiveness of the proposed fault diagnostic methods on Bently rotor test rig. The features in Time-domain and Frequency-domain are extracted from the data acquired in these tests.(2)Propose using Logistic regression to assess the machine running condition. Parameters for Logistic Regression Model are optimized though the method of cross validation. The value of probability corresponded to training data is confirmed approximate by fault feature.(3)Bring up using support vector machine model to predict the machine running condition. Basis function (RBF) for model is used. Analysis of influence on different basis function, model Parameters, prediction length. Use Mean Absolute Percentage Error and Root Mean Square Error to evaluate precision of the result about prediction.(4)Designed the framework based on SVM component with ActiveX control, according software technology. Wrote the DCOM using the ADO technology and VB language and created the fault diagnosis database using SQL.