The classical multi-objective particle swarm optimization algorithm (MOPSO) often faces issues due to its over-reliance on the global optimal solution (gBest) and the individual historical optimal solution (pBest). Its simplistic search logic and update paradigm reveal significant performance bottlenecks, especially when handling complex many-objective optimization problems (MaOPs). To address the shortcomings of MOPSO, This paper constructs a multi-objective and many-objective optimization framework called BIG-MaPSO. The algorithm constructs a multi-dimensional co-evolution mechanism. While maintaining the benefits of the traditional MOPSO for low-dimensional multi-objective problems, it creatively integrates the dual mechanisms of angle competition and Assisted evolution mechanism, which greatly improves the exploration efficiency of MaOPs. Test data indicates that BIG-MaPSO outperforms other algorithms on benchmark test sets, such as UF and DTLZ. Additionally, it has demonstrated its effectiveness in the complex practical application of neural architecture search.