This paper introduces a strong privacy-guaranteed BRNN query architecture based on PIR, and proposes an orthogonal optimization technique to further improve the query efficiency. Firstly, this paper analyzes the privacy-oriented BRNN query, secondly introduces the method of BRNN query processing, and finally proposes the orthogonal optimization technology to further improve the query efficiency and experimental verification. The key problems to be solved in this paper include improving the success rate of query, reducing the time of anonymous query and improving the accuracy of data modeling.