UAV Path Planning Using a State Transition Simulated Annealing Algorithm Based on Integrated Destruction Operators and Backward Learning Strategies

This study introduces a state transition simulated annealing algorithm that incorporates integrated destruction operators and backward learning strategies (DRSTASA) to address complex challenges in UAV path planning within multidimensional environments. UAV path planning is a critical optimization problem that requires smooth flight paths, obstacle avoidance, moderate angle changes, and minimized flight distance to conserve fuel and reduce travel time. Traditional algorithms often become trapped in local optima, preventing them from finding globally optimal solutions. DRSTASA improves global search capabilities by initializing the population with Latin hypercube sampling, combined with destruction operators and backward learning strategies. Testing on 23 benchmark functions demonstrates that the algorithm outperforms both traditional and advanced metaheuristic algorithms in solving single and multimodal problems. Furthermore, in eight engineering design optimization scenarios, DRSTASA exhibits superior performance compared to the STASA and SNS algorithms, highlighting the significant advantages of this method. DRSTASA is also successfully applied to UAV path planning, identifying optimal paths and proving the practical value of the algorithm.