Performance Evaluation and Optimization of Cold Tandem Rolling Thickness and Tension Control Based on Subspace LQG Benchmark

In the cold tandem rolling production process, thickness and tension control systems are essential. However, factors such as changes in production conditions and lack of maintenance can degrade the performance of these systems. To monitor the operation of controllers in real time, this paper introduces a performance monitoring algorithm based on the subspace Linear Quadratic Gaussian (LQG) benchmark. First, we use subspace identification algorithms to process historical rolling mill data to obtain subspace matrices and design a subspace-based LQG thickness-tension predictive controller performance evaluation benchmark. Next, the controller parameters of cold tandem rolling are optimized using particle swarm optimization, enhancing the overall system level. Finally, we evaluate the frame predictive controller online by solving real-time performance indicators of cold tandem rolling and comparing them with the established performance benchmark. Experiments prove that the subspace performance evaluation algorithm can accurately represent the performance level of the control system and effectively monitor controller performance changes.