An intelligent fault diagnosis method based on adaptive maximal margin tensor machine

Data driven intelligent method for fault diagnosis has become a widely used technology. However, the traditional machine learning methods are limited when using two-order signal feature for modeling. When the two-order signal features are vectorized as the input, their structure information will be lost. To fully protect the structural information, a tensor-based classification method is proposed, termed adaptive maximal margin tensor machine (AMMTM). The core of AMMTM is to establish a pair of interactive hyperplanes, so that each type of samples bounds nearing its corresponding hyperplane and away from another hyperplane as far as possible. Meanwhile, a deviation parameter is introduced into the 2-norm distance metric, which changes the constraints on the hyperplanes and maximizes the distance between the two hyperplanes, so as to improve the generalization ability and robustness. Two roller bearing datasets are used for validation, and the experimental results show that AMMTM has superior classification ability.