Model based fault detection for electrical drives with BLDC motor

Continuous improvements in microelectronic circuitry make possible the implementation of process diagnostics to a variety of systems in order to increase performances and reliability. The monitoring stage in the area of failure detection and isolation aims to classify and arrange the failure sources. In case of BLDC machines, one of the most important methods for failure detection and diagnosis starts with the analysis of the variations of the machine estimated parameters. The paper focuses on the implementation details of failure detection and diagnosis based on continuous time parameters estimation of BLDC motor mathematical model. Continuous time parameters are directly related to the physical characterizations of BLDC Motor. The parameters are estimated by the well-known prediction error methods devised for dynamic system identification. By changing the process coefficients and by applying statistical decision methods, failure detection occurs. The case of frequency domain identification in fail detection is also covered on a real CNC machine. Discrete Fourier Transform (DFT) with the particular case of Goertzel algorithms is implemented for fault detection purposes. Unlike existing work and results in the fault detection area, the failure can be detected among time-varying process parameters.