基于改进BP方法的精炼炉电压和电流软测量建模

Research on soft-sensor method for arc current and voltage of ladle furnace based on the improved BP method

在精炼过程中,精确地测量电弧电流和电弧电压对提高调节器性能、生产优质钢和提高冶炼效率有着重要的理论和实践意义。首先建立基于BP神经网络的电弧电流及电压软测量模型,然后针对BP神经网络收敛速度慢、易陷入局部极小值点的缺点,对模型进行改进,提出基于遗传算法的BP神经网络软测量模型。在MATLAB仿真平台中对建立的两个模型进行仿真比较,结果表明基于遗传算法的BP神经网络软测量模型在收敛速度、泛化能力等方面都要明显优于单一的基于BP神经网络建立的测量模型。

In the process of refining, the accurate measurement of arc current and voltage has theoretical and practical significance of improving the performance of the electrode regulator, producing high-quality steel and increasing the production efficiency. One soft-sensor model based on BP is established firstly. To deal with the defects of the steepest descent in slowly converging and easily immerging in partial minimum frequently of BP, the genetic algorithm is brought forward to solve the problem. Finally, two models are simulated under MATLAB for verifying their effectiveness and the simulation results are compared with each other. The result shows that the BP model based on genetic algorithm is much better than a single BP model on convergence rate and generalization ability.