Abstract: Tourism, as an important part of the tertiary industry, has shifted from the high-speed development stage to the high-quality development stage. In recent years, agricultural tourism has become the innovation point of supply-side structural reform in China. However, due to the short development time of agricultural sightseeing tourism in China, the existing evaluation system is not suitable for the evaluation of tourism competitiveness of agricultural sightseeing scenic spots, and the quality levels of scenic spots are not uniform. This study proposed an improved genetic neural network model to evaluate the tourism competitiveness of agricultural sightseeing scenic spots. The diversity of the population was improved by expanding the number of the population, and the two-stage solution was adopted to reduce the difficulty of solving the population, and then the genetic algorithm was processed in parallel. The improved algorithm improved the efficiency by about 30% and the performance by about 10%~20%. The parallel scheme proposed in this study and the competitiveness evaluation index constructed had high practicality and objectivity, which could provide some reference for related research.

Key words: agricultural tourism, tourism competitiveness, genetic neural networks, two-stage solution, parallelization