The optimization path of smart ice and snow sports tourism industry development under hybrid neural network model

With the swift development of the smart ice and snow sports tourism industry, how to scientifically predict tourist flow and optimize resource allocation has become an urgent problem to be solved in the industry. However, most existing studies rely on a single neural network model, which makes it difficult to balance time-series and static features. In addition, there is a problem of local optimality in parameter tuning, which affects prediction accuracy and model stability. In order to address this problem, this study proposes a Hybrid Neural Network (HNN) model based on a Multi-Layer Perceptron (MLP) and Long Short-Term Memory (LSTM). Meanwhile, it introduces the Chaotic Adaptive Particle Swarm Optimization (CAPSO) algorithm to optimize the network parameters and hyperparameters. This model can simultaneously capture the temporal dependence of historical tourist flow and multi-dimensional static features, thereby improving prediction accuracy and generalization ability. Experimental verification is conducted using public datasets and actual ski resort data. The results show that for tourist number prediction, the Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and R-Square (R²) of the CAPSO-MLP-LSTM model are 108.4, 85.3, and 0.981; for room occupancy rate prediction, its RMSE, MAE, and R² are 0.031, 0.027, and 0.935. These performance indicators are significantly better than those of traditional models such as LSTM. The research results can provide decision support for the ski resorts' operation management, resource scheduling, and marketing strategies. Meanwhile, these findings can offer references for the development planning and management optimization of the smart ice and snow sports tourism industry.