A fast fuzzy stator condition monitoring algorithm using FPGA

Abstract

This paper presents a new methodology for realtime detection of stator faults in induction motors. It is based on the evaluation of the magnitudes of three phase current signals. The severity of fault is indicated by designed fuzzy system. The algorithm consists of two stages: feature extraction and classification. All acquired current signals are based on IEEE-754 single precision floating point format. In the feature extraction stage, floating point based features were obtained and then converted to the integer format. In the classification stage, all coding are written in VHDL using integer data type format. The proposed method classifies the stator related faults at higher level speed with better accuracy than conventional offline method. This is the first time that FPGA based fuzzy system has been employed to induction motor faults. With the proposed method, it is clearly identified the motor condition as either healthy or damaged, allowing one to perform a simple real time detection. A light computation load is required in order to perform the calculation of the algorithm. In addition, the implemented FPGA design allows the system reconfiguration to attain specific measurement conditions on different types of motors without hardware modifications.

FPGA-Embedded Anomaly Detection System for Milling Process

“…In [16] the first full hardwarebased implementation of the contracting autoencoder [17], comprising hardware-implemented learning was provided. A methodology for real-time stator condition monitoring of an induction motor using a fuzzy system implemented on FPGA was presented in [18]. The fuzzy system designed in VHDL was successfully compiled, and simulated and the results obtained from FPGA were compared with the results obtained from the Matlab Fuzzy Logic Toolbox.…”