Abstract: series of subtasks，resulting in high viewpoint position errors，proposes an intelligent assisted unmanned aerial vehicle inspection path automatic optimization method based on an LLM. Through Swarm，high-level instructions are transformed into inspection path optimization steps to complete task allocation and generate the LLM inspection task planning framework. Combine the planning chain to complete task decomposition，construct inspection path task planning prompt engineering, and complete logical task decomposition；generate the cost function for automatic optimization of the inspection path and the obstacle restriction space for inspection. Based on satisfying the yaw and pitch angles，construct an automatic optimization model for the drone inspection path. The experimental results show that after applying this method，the positive/negative viewpoint errors of the inspection path under different viewpoint coordinates are always low，and there is no significant optimization fluctuation，indicating that this method has high reliability.