Work

Master–Slave Agricultural Machinery Cooperative Harvesting Control Based on VMD-Transformer-LSTM and Dual-Layer MPC

Year: 2025

Type: article

Abstract: During field cooperative harvesting operations, the accuracy of the tracking behavior between the master and slave unmanned agricultural machines has always been a key factor affecting the quality of cooperative operations. To address this issue, this paper proposes a cooperative harvesting control method based on the prediction model combined with a dual-layer model predictive control (MPC). First, the variational mode decomposition (VMD) algorithm is used to decompose the historical speed, acceleration, and relative distance between master and slave machines into several intrinsic mode functions (IMF)with different frequencies. Then, the Transformer-LSTM model is employed to predict the future speed sequence of the master machine. Based on this, the future speed sequence is input into the dual-layer MPC. The upper-layer MPC adjusts the slave machine’s speed to approach the master machine’s speed and outputs a reference speed signal to the lower-layer MPC. The lower-layer MPC aims to minimize the deviation of the relative distance between the master and slave machines. Finally, the output is the final slave machine speed control signal. Experiments on master machine speed prediction, slave machine tracking, and cooperative harvesting operations were conducted. In the master machine speed prediction experiment, the VMD-Transformer-LSTM model showed significant performance advantages compared to the traditional LSTM, Transformer, and Transformer-LSTM models. The results of the slave machine tracking experiment indicated that the distance deviation in straight-line tracking was controlled within 7.1 cm, while the distance deviation in steering tracking was controlled within 13.7 cm, significantly improving the tracking accuracy. When using the proposed method for cooperative harvesting operations, the non-operating time was reduced by 58.62%, and the harvesting efficiency increased by 33.74%. This provides technical support for multi-machine cooperative harvesting.

Source: Agronomy

DOI: 10.3390/agronomy15102308