Abstract: Particle-In-cell (PIC) is a numerical algorithm that combines particle simulation and grid methods, and is widely used to simulate the behavior of charged particles in electromagnetic fields. Due to the need to track the movement of a large number of particles over time, it consumes a huge amount of computing resources and incurs high time costs. This paper proposes a low-rank method using particle blocks as the evolution units. By constructing a particle matrix composed of a large number of discrete particles and applying Singular Value Decomposition (SVD) to extract the main modes of the particle blocks, a low-rank approximation form corresponding to the collective particles is obtained, and a time integrator based on dynamic low-rank approximation method is used to update their motion states. The research results indicate that during the time evolution process, the low-rank particle blocks effectively preserve the continuity and dynamic consistency of physical distributions. Errors accumulate gradually within the predefined loss tolerance threshold, ensuring robust long-term simulation stability. The low-rank form studied in this paper not only has the ability of data compression and accelerated computation, but can also explore kinetic effects from the perspective of collective particle modes, providing a new technical path for the study of plasma kinetic effects.