The issue of cyber security for cyber-physical systems (CPSs) has become increasingly prominent in recent years. To ensure the secure operation of CPSs and the reliability of data, this study investigates the problem of distributed multi-level elastic state estimation for unmanned surface vessels (USVs) under denial-of-service (DoS) attacks and false data injection (FDI) attacks. First, a privacy-preserving model based on stochastic encryption technology is proposed, and the impact of the attack on the estimation error and system estimation performance is analyzed under different scenarios of the joint FDI and DoS attacks, and FDI dual-channel attacks. Then, to defend against FDI attacks, data is encrypted throughout the communication process. To counter DoS attacks, an encryption compensation mechanism is introduced. The open network environment exposes USVs to the risk of combining attacks during data transmission. These attacks threaten the authenticity of the states of the system's state. To address this, a distributed multi-level elastic state system is constructed to resist combined attacks and ensure the acquisition of the actual states of the CPS. Finally, through simulation and comparative experiments on the velocities states of USVs, the effectiveness of the proposed methods is validated.