Supervised fault learning using rule-generated samples for machine condition monitoring

摘要

A machine fault diagnosis system is provided. The system combines a rule-based predictive maintenance strategy with a machine learning system. A simple set of rules defined manually by human experts is used to generate artificial training feature vectors to portray machine fault conditions for which only a few real data points are available. Those artificial training feature vectors are combined with real training feature vectors and the combined set is used to train a supervised pattern recognition algorithm such as support vector machines. The resulting decision boundary closely approximates the underlying real separation boundary between the fault and normal conditions.

主权项

1. A method for classifying a measured test feature vector as representing one of a normal machine condition and a fault machine condition, the measured test feature vector including a set of feature states relating to a machine at a particular time, the method comprising:
receiving a manually defined rule establishing a set of feature state ranges indicating the fault machine condition; using probability distributions over the feature state ranges indicating the fault machine condition to sample feature state ranges established by the manually defined rule, generating a set of artificial sample feature vectors independently of a training set of measured training feature vectors, each artificial sample feature vector including an annotation indicating the fault machine condition; annotating the training set of measured training feature vectors by assigning an annotation to each measured training feature vector of the training set, the annotation indicating one of the normal machine condition and the fault machine condition; training a supervised pattern recognition algorithm using an enhanced training set comprising the training set of measured training feature vectors and the set of artificial sample feature vectors; and classifying the measured test feature vector using the trained supervised pattern recognition algorithm.