With the advancement of underwater robotics technology, traditional methods for performance testing and evaluation of underwater equipment are no longer sufficient to meet the diverse application needs of intelligent submersibles, especially Autonomous Underwater Vehicle (AUV). Building upon the performance testing of traditional underwater equipment, the testing of intelligent submersibles should more closely align with user requirements, taking into full consideration the actual operational tasks such as marine scientific surveys and marine archaeological searches that impose specific testing demands on these submersibles. Focusing on the core application functions of intelligent navigation and intelligent detection, this study analyzes and researches typical operational tasks in marine scientific investigation and archaeological searches. It proposes a task-based scenario planning for the performance testing of intelligent submersibles, providing a reference for constructing performance evaluation methods that serve operational tasks.