In order to solve the problem of heterogeneous data sources and data sharing in information systems, a virtual integration framework of multi-source heterogeneous data is proposed. Since GAV (Global-As-View) pattern mapping method in data integration system is less efficient when faced with uneven distribution of information, GAV method is improved, and the pattern mapping method based on HGAV (Hierarchical-Global-As-View) is proposed. By introducing the intermediate data source pattern, a hierarchical global view is formed, which greatly reduces the mapping space. In this way, the mapping set is simplified, and the query is easier to rewrite and optimize. The proposed algorithm is verified by the five main types of data in the Ningdong Intelligent Environment Protection project. The experimental results show that the pattern mapping algorithm based on HGAV improves the efficiency of data integration and shortens the query time, compared to the GAV schema mapping algorithm.