Abstract: Accurate tracking control of an unmanned roller is of great significance to ensuring compaction quality. However, high-core rockfill dams feature a wide range of the particle size of dam materials, an uneven thickness of the paving layer, and a great heterogeneity of the roller working surface; all these factors lead to large dynamic disturbances to roller control. The traditional proportional-integral-derivative (PID) control algorithm uses fixed parameters for deviation correction, and confronts difficulties in timely correction of the tracking deviation caused by dynamic disturbances. Aimed at this issue, this paper uses the butterfly optimization algorithm (BOA) to dynamically optimize the PID control parameters and develops a BOA-PID unmanned roller tracking control method. A body tilt model is constructed to correct the positioning error of the roller under the condition of an uneven working surface; the correction distance is dynamically predicted based on a kinematic model. Then, by taking the minimized correction distance as the goal function, PID is dynamically optimized by BOA including its proportional, integral and differential parameters. Finally, a steering control value for the roller is calculated using the optimized PID, so as to overcome the dynamic disturbance under the complex condition of high-core rockfill dams. This paper gives a case study of the Lianghekou Dam－a large-scale water conservancy and hydropower project in Southwest China－where we have conducted field experiments to verify the effectiveness of this BOA-PID control method. Results show that its correction capability is better than that of GA-PID (genetic algorithm), PSO-PID (particle swarm optimization), DA-PID (dragonfly algorithm), or traditional PID. And it can realize an accurate tracking control of the unmanned roller under the complex condition of high-core wall rockfill dam construction, and reduce the average tracking error of the rockfill material (4.44 cm) and the core wall material (3.32 cm) to a level below 5 cm.

Key words: high core rockfill dam, unmanned roller compactor, tracking control, dynamic disturbance, butterfly optimization algorithm (BOA), proportional-integral-derivative (PID) control