Research on Control Strategy for Weighing Process of Ladle Furnace

Due to the uncertainties in the alloy weighing process, traditional control strategies struggle to meet the increasing accuracy requirements. This paper proposes a weight-flow dual-loop learning control strategy. The outer loop, weight control, utilizes an iterative self-learning algorithm to optimize the predefined S-shaped feeding flow curve, thereby mitigating the impact of system uncertainties and disturbances between batches on the final feeding error. The inner loop, flow control, employs a reinforcement learning PID algorithm to track the optimized flow curve obtained from the outer loop. It dynamically adjusts the control strategy through real-time interaction with the system to overcome within-batch disturbances and adapt to time-varying characteristics. Numerical simulations are conducted to verify the effectiveness, accuracy, and robustness of this control strategy.