Due to the polymorphic uncertainties in microgrids (MGs), prohibitive computational burden is produced in reliability assessment. In this work, a novel sequential sampling algorithm (NSSA) compatible with sequential Monte Carlo (SMC) simulation is developed to overcome the computational burden. First, optimal probability density functions (PDFs) of random variables are worked out based on variation method. Then, optimal PDFs are employed to chronologically simulate the random states of microturbine (MT), photovoltaics (PV) and time varying load with improved computational efficiency. Therefore, the convergence of reliability assessment is accelerated accordingly. A series of case studies have been conducted, and the computational results show that NSSA provides a favorable sampling efficiency and adaptability to system conditions in reliability assessment of MGs. At last, based on optimal PDFs produced by NSSA, dominant joint PDF (DJ-PDF) is defined and employed to quantify the contributions of different scenarios to the reliability indices. Case studies have confirmed that DJ-PDF can provide detailed information for scenario-based reliability analysis.
INDEX TERMSMicrogrid, sequential sampling, reliability assessment, computational efficiency, coefficient of variation NOMENCLATURE Random Variables u random variable sampled for random operation hours v random variable sampled for random repairing hours q random variable defined to model time varying load p random variable defined to model intermittency of PV Indices and Sets f p estimate of fp * (p) on the wp-th subinterval of [0, 1] FR(•) reliability test function in terms of u, v, q and p with uniform PDFs FR ' (•) reliability test function in terms of u, v, q and p with varying PDFs GR(•) reliability test function in terms of the hourly states of MT, PV and load FL[•] Functional in terms of fu(u), fv(v), fq(q) and fp(p) I. INTRODUCTION