Optimization modeling and verification of building materials cost budget for urban rail transit construction

This study addresses the challenges inherent in the management of construction costs within the urban rail transit sector. Utilizing recent engineering data from the Shenzhen Metro Group, the research has established an "1 + 3 + N" intelligent management framework and has innovatively developed a multi-modal timing prediction algorithm. This algorithm synergizes the Prophet cycle model with the LSTM deep learning network, integrating via the Attention mechanism. This integration notably enhances the precision of construction material price forecasts. For the analysis of ready-mixed concrete price data in Guangzhou and Shenzhen from June 2023 to March 2024, the study employed the hybrid Prophet-LSTM-Attention model, which was further combined with the Supply Chain Elasticity Index (SEI) and GARCH methodologies to conduct a quantitative analysis of supply chain risks and predict prices. Empirical outcomes indicate that the hybrid model exhibited optimal predictive performance, achieving an approximate reduction in prediction error by 25-30%. We also embed Prophet-LSTM-Attention model into the existing BIM(Building Information Modeling) platform of procurement system to generate building materials price warning and procurement suggestions in real time. Experimental results show that the model has good generalization ability. This research overcomes the constraints of conventional segmented management approaches. A hybrid Prophet-LSTM-Attention model is developed to optimize the cost prediction of building materials and supply chain risk analysis in urban rail transit construction, evaluate the environmental benefits and quantify the ability of resource optimization according to the forecast results, and put forward the path for green supply chain decision making.and improve industry decision-making and cost efficiency.