Abstract: To solve the problems of low robot inspection efficiency and lack of obstacle avoidance ability in current power system,an Improved Marine Predator Algorithm (IMPA) for obstacle avoidance was proposed. Firstly,the spiral complex path search strategy based on the Archimedean spiral curve was used to expand the global search range,thus improving the stability of the algorithm. Secondly,the nonlinear convex decreasing weight was introduced to balance the global and local ability of the algorithm,so as to realize the dynamic update of predator and prey population position. In addition,the golden sinusoidal algorithm was combined to update the location of prey,so as to reduce the search range of predator population,and improve the precision and speed of convergence. Finally,experiments were carried out in the distribution room and substation environment. The results show that compared with the basic MPA,the improved ant colony algorithm and the hybrid MPA,the IMPA is more dominant in the travel distance and time,it is more conducive to improve the robotic autonomous obstacle avoidance ability and operational efficiency.

Key words: intelligent power, electric inspection robot, obstacle avoidance, marine predator algorithm