Abstract: In order to simultaneously balance privacy protection and query performance, a privacy protection query strategy based on adaptive combination filters is proposed. An adaptive window locally sensitive hash filter was introduced, which combines with Bloom filter to perform fuzzy matching for similar or within-range values. An improved local differential privacy mechanism was proposed to improve query efficiency for stream data with similar or lower privacy budget and reduce real-time query response time. Experiments on real and synthetic English word datasets verify the effectiveness of the proposed method.