Abstract: As maritime networks continue to expand, The compute-intensive tasks of offshore monitoring terminals are growing exponentially. Due to the limitations of their own communication and computing resources, it is difficult to handle the increasingly diverse business needs. In the limited budget uplink offloading scenario, Traditional ocean communication models find it difficult to improve the system's total computational workload while meeting energy, efficiency, and cost requirements.Therefore, this paper introduces a Double-Intelligent Reflecting Surface (IRS) collaborative architecture, which deploys two distributed IRS relay collaborative users to offload tasks to the onshore MEC server. This paper considers the joint optimization of base station receiving beamforming, dual IRS joint phase shift matrix, user transmission power, and CPU computing frequency and designs a joint task offloading and resource allocation algorithm to maximize the total computational task of the system under communication and computing resource constraints. This non-convex optimization problem is decomposed by introducing the block coordinate descent (BCD) idea and solved by an efficient alternating optimization algorithm based on maximal ratio combining (MRC), Lagrange multiplier method and bisection search. The simulation results show that the proposed dual IRS assisted offshore collaborative offloading scheme improves the system's performance., 随着海事网络的不断扩展，海上监控终端的计算密集型任务呈指数级增长。由于自身通信和计算资源的限制，很难处理日益多样化的业务需求。在有限预算的上行卸载场景下，传统的海洋通信模型难以在满足能量、效率和成本要求的同时提高系统的总计算量，为此提出了一种双智能反射面（IRS）协作架构，部署两个分布式IRS中继协作用户，将任务卸载到岸上MEC服务器上.考虑基站接收波束形成、双IRS联合相移矩阵、用户发射功率和CPU计算频率的联合优化，设计了一种联合任务卸载和资源分配算法，在通信和计算资源约束下最大化系统的总计算任务。通过引入块坐标下降（BCD）思想，将非凸优化问题分解，并采用基于最大比合并（MRC）、拉格朗日乘子法和对分搜索的交替优化算法求解。仿真结果表明，提出的双IRS辅助海上协同卸载方案提高了系统的性能。, 随着海事网络的不断扩展，海上监控终端的计算密集型任务呈指数级增长。由于自身通信和计算资源的限制，很难处理日益多样化的业务需求。在有限预算的上行卸载场景下，传统的海洋通信模型难以在满足能量、效率和成本要求的同时提高系统的总计算量，为此提出了一种双智能反射面（IRS）协作架构，部署两个分布式IRS中继协作用户，将任务卸载到岸上MEC服务器上.考虑基站接收波束形成、双IRS联合相移矩阵、用户发射功率和CPU计算频率的联合优化，设计了一种联合任务卸载和资源分配算法，在通信和计算资源约束下最大化系统的总计算任务。通过引入块坐标下降（BCD）思想，将非凸优化问题分解，并采用基于最大比合并（MRC）、拉