Nowadays, due to the great advent of sensor technology, the data of all appliances in a house can be collected easily. However, with a huge amount of appliance usage log data, it is not an easy task for residents to visualize how the appliances are used. Mining algorithms is necessary to discover appliance usage patterns that capture representative usage behavior of appliances. If some of our representative patterns of appliance electricity usages are available, we may be able to adapt our usage behaviors to conserve the energy easily. In this paper, we introduce (i) two types of usage patterns which capture the representative usage behaviors of appliances in a smart home environment and (ii) the corresponding algorithms for discovering usage patterns efficiently. Finally, we apply our algorithms on a real-world dataset to show the practicability of usage pattern mining.