<p>This article combines Archard wear algorithm with Abaqus finite element software for wear simulation and verifies the accuracy of the algorithm through experiments, with an error of 1.6%. This method was used to conduct an in-depth analysis of the wear of the three-piston caliper brake pads, in order to explore their braking performance. On this basis, sensitivity analysis was conducted on the three-piston layout structure using Taguchi method, and a genetic algorithm optimized neural network surrogate model (GA-BP) was constructed. The non-dominated sorting genetic algorithm II (NAGA-II) was introduced for multi-objective optimization to obtain the Pareto curve. The research results show that the contact pressure distribution of the three pistons is uniform, and the maximum values of contact stress and pressure are concentrated at the groove. Observing the Pareto surface, it can be seen that there is a contradiction between wear quality and contact area. While reducing wear, the overall contact area between the front and rear parts of the brake pad shows a downward trend; by using multi-objective optimization methods, the wear amount of brake pads was reduced from 2778.3 to 2731.91&#xa0;mg, and the contact area was increased from 2533 to 2779.3&#xa0;mm². The multi-objective optimization method used in this article has shortened the cycle and cost of optimizing calipers, which has a positive significance for the design of calipers.</p>