Improved BP Neural Network Database for Plate-Type Lifting Lug Design

Utilizing the nonlinear fitting capability of particle swarm optimization algorithm for the initial database of plate type lifting ear design. Divided into nearly 3000 sets of data based on different lifting angles (from 0 to 30), lifting weights (from 1 ton to 60 tons), and different materials (Q235 and Q345). Using particle swarm optimization algorithm and penalty function, with minimum weight as the objective function, obtain the optimal solution for each group to form a database. Use this database as the initial database to train an improved BP neural network and obtain the trained weights and thresholds.