The construction of a power grillage is of great significance for promoting local economic development. Identifying the characteristics of foundation damage is a prerequisite for ensuring the normal service of the power grillage. To investigate the bearing mechanism and failure mode of the grillage root foundations, a novel research method with a transparent soil material was used to conduct model tests on different types of foundations using particle image velocimetry (PIV) technology. The results indicate that, compared to traditional foundations, the uplift and horizontal bearing capacities of grillage root foundations increased by 34.35% to 38.89% and by 10.76% to 14.29%, respectively. Furthermore, increasing the base plate size and burial depth can further enhance the extent of the soil displacement field. Additionally, PIV analysis revealed that the roots improve pile-soil interactions, transferring the load to the surrounding undisturbed soil and creating a parabolic displacement field during the uplift process, which significantly suppresses foundation displacement. Lastly, based on experimental data, an Elman neural network was employed to construct a load-bearing capacity prediction model, which was optimized using genetic algorithms (GAs) and the whale optimization algorithm (WOA), maintaining a prediction error within 3%. This research demonstrates that root arrangement enhances the bearing capacity and stability of foundations, while optimized neural networks can accurately predict the bearing capacity of grillage root foundations, thus broadening the application scope of transparent soil materials and offering novel insights into the application of artificial intelligence technology in geotechnical engineering. For stakeholders in the bearing manufacturing industry, this study provides important insights on how to improve load-bearing capacity and stability through the optimization of the basic design, which can help reduce material costs and construction challenges, and enhance the reliability of power grillage infrastructure.

The new type of support disc-type anchor is an expanded body anchor with broad application prospects, and its load-bearing performance is significantly better than that of traditional anchors. However, there is a problem of premature shear damage in traditional support disc-type anchors. In order to solve this problem, this paper improves the traditional support disk anchor. It conducts cyclic loading tests on the new type of support disc-type anchors with different support disc diameters, support disc thicknesses, anchoring diameters, and anchoring lengths so as to simulate the repeated loads that the anchors are subjected to in actual projects. The function model suitable for predicting the bearing capacity of the new type of support disc-type anchors was derived by nonlinear fitting of some data using the function model and verified by comparing it with the measured data. A functional model predicts the bearing capacity of the new type of support disc-type anchors through nonlinear fitting of the data, validating the model against measured results. The study reveals that factors such as support disc diameter and anchoring length have the most significant impact on pullout bearing capacity. In contrast, increasing the anchoring diameter and shortening the anchoring length may lower the pullout bearing capacity. The Q-s curve divides into five stages, where the lateral friction force between the anchoring and the soil, along with the bulb resistance at the supported disc, collectively generates the bearing capacity of the new type of support disc-type anchor. The Belehradek function model proves most effective in describing the Q-s curve for these anchors during testing, demonstrating high accuracy and strong engineering practicality.