A fuzzy logic based online energy management system (FLEMS) is designed in this paper to achieve the optimal electricity cost in a residential Microgrid (MG). The proposed FLEMS is combined by a local energy price model (LEPM) and a fuzzy-logic strategy. The LEPM will preprocess the sampling data to estimate the electricity market and local MG status. The fuzzy-logic mimics the artificial intelligent assessment to economic issues and make decision for the charging and discharging operation for energy storage system (ESS). In the FLEMS, not only electricity price and supply-demand balance, but also ESS state of charge are considered for the efficient and stable operations. The proposed method does not relay on the accurate prediction of renewable energy source and local loads. Historical experience of the system is involved by the LEPM and guides the ESS operation in the fuzzy-logic. A real-world data based household-level residential MG model is established to validate the performance of the FLEMS. A hourly-resolution-Particle swarm optimization (PSO) with perfect day-ahead prediction is implemented as the baseline to verify the superiority of the proposed method.