Cost-effective on-line fault diagnosis of induction motors using the reduced modulus of the current park's vector

Abstract

This paper presents a new algorithm for the practical implementation on embedded devices of the extended current Park's vector approach (EPVA), which substantially reduces the hardware and computational requirements needed for applying this technique. The EPVA is a proven technique for fault diagnosis of induction motors working in steady state regime. However, this approach has some practical limitations: it requires the simultaneous measurement of three-phase currents, using three high-resolution, synchronous analog-to-digital converters, operated at high sampling rates to avoid aliasing artifacts, and during long periods for reaching suitable frequency resolution; thus, a huge amount of data must be stored and processed. To solve these problems, a cost-effective implementation of the EPVA is presented in this paper. The proposed approach only requires a single ADC channel sampled at a very low frequency rate and, besides, it reduces the computational requirements (storage capacity, calculation time) to less than 1% of those required by the conventional EPVA. In this way it can be implemented in real time, using low cost embedded devices. This approach is theoretically introduced and experimentally validated using commercial induction motors with broken bars, under different supply and load conditions.