Fault Detection in Brushless DC Motor via Particle Filter

In this article, the brushless dc motor (BLDC) fault detection problem is adressed. The fault studied is the power supply interruption in one of its stator phases. This problem has already been solved with different parameter estimation techniques, such as the Least Squares Estimator and the Recursive Least Squares Estimator, in which the parameters of a linear model related to the fault are estimated, allowing the fault detection. However, in this paper the solution is based on augmenting the model state vector with a parameter that indicates the fault and on the estimation of the system states with a nonlinear state observer, which is the particle filter. This method was applied to detect the fault of a real BLDC motor and the results were compared to the values obtained with the least squares method. We conclude that the experimental results validate this method.