The development of the National Park System (NPS) in the United States leads the construction of natural protected areas in the world. The composition, characteristics and trends of its scientific research have very important reference significance for the research on protected areas and the construction of national park system and science in China. A total of 10,954 related scientific papers were collected from six databases, and combined with TF-IDF and LDA text analysis algorithms, the system composition and evolution trend of scientific research on national parks in the United States in the past 150 years were studied. The results show that: ①The structure of NPS scientific research system has formed the overall framework of "four beams and eight pillars". "Four beams" refer to the four focuses of science, which are plants, animals, natural environment and human behavior, and "eight pillars" refer to the eight research areas and 23 research directions, including public administration, recreation, geology and soil.②The 18 feature words showed that the popularity of NPS research showed great difference, and the popularity of most feature words showed an upward trend. 8 feature words such as land and management are in the high heat range, 6 feature words such as animal and plant are in the middle heat range, and 4 feature words such as fire and visitor are in the low heat range. Land and management are ongoing topics in NPS research, basic research represented by animals and plants has become relatively mature, and emerging fields represented by climate change and diversity may become hot topics in the future. ③ In the initial stage of NPS system, the research topics are around the resource survey and the driving factors for the establishment of the system. During the improvement period of NPS, the research topics focus on infrastructure and service facilities construction; During the period of stagnation and redevelopment of NPS, research topics revolve around the Great Depression, World War II and other major events; In recent years, the research topics are multidisciplinary and begin to develop plurally based on ecology. In the context of big data, the combination of TF-IDF and LDA and other analysis algorithms may become a new way to reveal the inner relationship and evolution characteristics of the same object and different disciplines.