A New Binary Classification Twin Uncertain Support Vector Machine with Imprecise Data

In the recent years, more and more people choose high-speed railway as the transportation to travel. Therefore, ensure the safety of passengers is of great importance. The braking system is a vital system to ensure the safety of railway, therefore, there are many methods focusing on the fault diagnostics of the braking system. Because of the errors of sensors, the data observed from the braking system are not precise. Besides, due to the continuous operation of the railway, the parameters are changing all the time, therefore, it is difficult to obtain the precise data. Based on the above reasons, the method to deal with imprecise data is important for us to find the faults of the braking system. Twin support vector machine (TWSVM) is a commonly used model to classify the normal data and fault data in the field of fault diagnostics, which is the extended model of the support vector machine (SVM). However, the traditional TWSVM method cannot deal with the imprecise data. Considering the imprecise data as the uncertain variable in Uncertainty theory, twin uncertain support vector machine (TWUSVM) is proposed in this paper to classify the imprecise data. It is proved by a real dataset of high-speed railway braking system and two puiblic datasets that the proposed model has good performance in classifying the imprecise data. Therefore, the proposed method can be widely applied in the field of fault diagnosis.