Despite extensive research on slope seepage mechanisms, a reliable long-term prediction method for slope deformation considering rainfall remains undeveloped, largely due to the complexity of rainfall-induced slope instability. This study leverages a project in slope engineering to explore slope deformation under heavy rainfall using intelligent monitoring techniques and genetic algorithm (GA) optimization for neural network prediction. By analyzing slope deformation patterns under varied rainfall intensities, results reveal that limited rainfall has minimal impact on slope stability, whereas excessive rainfall disrupts internal seepage patterns, increasing pore water pressure and reducing soil shear strength, it thereby enhances the risk of slope instability and potential landslides, significantly impacting slope stability. The GA-optimized network accurately captures abrupt slope deformation stages, avoids local optima, and provides a viable framework for early warning of slope instability.

The jacket substructure is a critical component of the offshore wind turbine (OWT) that is the interface between the transition piece at the top and the grouted connection. This paper presents a comprehensive study on the optimization of a jacket substructure to achieve greater cost efficiency while maintain acceptable structural performance. A fast parametric finite element modelling (FEM) approach for jacket substructures was firstly proposed. The generated models took into account realistic loading conditions, including self-weight, wind load and sectiondependent wave load, and soil-pile interaction. Parametric studies were conducted afterwards to investigate the trends of the mass and response of the jacket substructure with respect to the variation of geometric and sectional parameters. Optimizations of the jacket substructure were carried out using parametric optimization and numerical genetic algorithm (GA) optimization under three different optimization strategies corresponding to three groups of objective and constraint functions. The trends obtained by parametric analysis were used to guide the parameter selection in parametric optimization, while a rank-based mutation GA was established with the proposed efficient FEM embedded in as the solver to the optimization objective and constraint functions. Parametric optimization gained its advantage in computational efficiency, and the mass reduction were 6.2%, 10% and 14.8% for the three strategies respectively. GA optimization was more aggressive as the mass reductions were 16.8%, 22.3% and 34.3% for the three strategies, but was relatively more computational intense. The two proposed optimization methods and the three optimization strategies are both expected to be applied in practical engineering design of OWT jacket substructure with good optimization output and high computational efficiency.