With the power system's development and intelligence improvement, the demand for appliance identification technology is increasing. Appliance identification refers to analyzing and processing appliance data in the power system to obtain detailed information and signatures. However, traditional appliance identification methods usually require surveying users or deploying additional sensors. This intrusive method has problems such as high costs and difficulty in data acquisition. To overcome that problem, we propose a non-intrusive load monitoring (NILM) method based on a hash image retrieval algorithm. First, we represent the appliance data of the power system as an image and convert it into a compact and unique hash code using a hashed image retrieval algorithm. Then, we built a hash image library containing hash images of various appliance types and their corresponding appliance feature information. When appliance identification is required, we only need to compare the appliance V-I trajectory image to be identified with the hash image in the library and determine the most matching appliance type by calculating the similarity between the hash codes. The experimental results show that the NILM method based on the hash image retrieval algorithm is efficient and accurate. Therefore, the proposed method has broad prospects and potential in practical applications and can provide strong support for the operation and management of power systems.