Abstract: Aiming at the problem of motor bearing failure identification problems, a fault diagnosis method for motor bearings based on BAS(Beetle Antennae Search) algorithm and PNN( Probabilistic Neural Network) is proposed. The LLE (Locally Linear Embedding) algorithm is used to obtain the sensitive characteristics of the vibration signal to ensure the reliability and sensitivity of the vibration signal. The Beetle Antennae Search algorithm is used to find the optimal smoothing factor in the model to avoid the influence of subjective empirical selection on the diagnosis results. The experimental results show the effectiveness of the method, and it can accurately identify the fault type.

Key words: motor bearing, fault diagnosis, beetle antennae search algorithm, probabilistic neural network (PNN), locally linear embedding (LLE)