Fault diagnosis in switched‐linear systems by emulation of behavioral models on FPGA: A case study of current‐mode buck converter

Abstract

AbstractThis paper presents an Field‐Programmable Gate Array (FPGA)‐based fault emulation and detection framework for switched‐linear systems commonly encountered in power electronics. The proposed framework exploits high‐level behavioral modeling that maps analog functionality into fixed‐point architectures supported by FPGAs. This work is mainly concerned with modeling and validation of effectiveness of a fault diagnoser of a power converter much before its fabrication. The proposed concept of fault diagnosis at presilicon stage overcomes several reliability issues in power electronic systems and enables design and development of test resources, verification of test bed, and interconnecting circuitry at an early stage of the design. To illustrate applicability of this environment, well‐accepted fault events including hard and soft faults are emulated with the help of a nonideal model of a buck converter and a piecewise linear detection filter, which follow an observer‐based residual generation for fault detection. The paper also shows that the proposed method is equally applicable for detection of faults under unstable operation of the converter. The effectiveness of the proposed framework is evaluated through FPGA‐based simulation (or “emulation”) of a peak‐current‐mode–controlled buck‐type switching converter implemented on Xilinx Kintex‐7 FPGA in MATLAB/Simulink environment with Xilinx System Generator support. To the best of our knowledge, fault diagnosis in power converter by emulation of its nonideal model on FPGA is attempted for the first time in the literature. Experimental results show that power converter faults can be easily detected by monitoring detection filter residuals while some of the faults can be precisely identified.