INTELLIGENT PREDICTION OF ANCHORAGE FORCE FOR EARTHEN ANCIENT HERITAGE SITES BASED ON IMPROVED PSO-BP NEURAL NETWORK

The “minimum intervention” principle of ancient architecture strictly prohibits large-scale in situ testing during reinforcement design, resulting in anchoring design often having a greater empirical and random nature. In recent years, the advantages of artificial intelligence in data mining, high efficiency, and precision have provided a new perspective for the protection of ancient architecture. How to coordinate the "minimum intervention" principle and the scientificization of reinforcement design has become an important topic in the intelligentization of ancient architecture protection. Therefore, this paper introduces an adaptive inertia weight and asymmetric learning factor to improve the traditional particle swarm optimization algorithm, and then optimizes the initial weight and threshold of the BP neural network to construct a novel particle swarm optimization-backpropagation (IPSO-BP) anchorage force intelligent prediction model. Taking carbon fiber bamboo anchors as an example, the paper combines in situ and model tests to establish a sample data set of anchorage forces, considering factors such as anchor length, diameter, inclination angle, grout strength, hole diameter, and carbon fiber wrapping interval. The data learning and prediction results show that the IPSO-BP model has better robustness, efficiency, and accuracy, with a root mean square error and average absolute error that are respectively reduced by 61.3% and 31.9% compared with the traditional PSO-BP model. Based on Spearman's rank correlation coefficient theory, the paper further analyzes the sensitivity of anchorage force to different influencing factors, the results of which show that anchor length is the key factor affecting anchorage force, while the drilling volume will directly affect the degree of damage to the soil site during anchoring construction. Subsequently, the paper takes anchor length and hole diameter as the design variables and conducts single-objective and multi-objective optimization analysis to obtain the optimal design scheme that maximizes anchorage force and minimizes drilling volume. The research results can provide technical support and theoretical reference for the intelligent development of soil site reinforcement and protection.