考虑间歇过程重复特性的状态估计方法

A State Estimation Method Based on Repetitive Nature of Batch Process

考虑间歇过程的重复特性,文章提出一种基于粒子滤波的迭代状态估计算法。通过该算法迭代先前批次的状态估计值,为当前批次提供一种测量值信息,从而利用先前批次的数据信息来提高当前批次的估计精度。与先前批次相比,当前批次的初始状态误差随着批次数的增加逐步减小,趋于收敛,且整个批次的估计效果得到提高。最后,分别以数字仿真和啤酒发酵过程为例,验证了状态估计算法的有效性。

Considering the repetitive nature of batch processes,an iterative learning strategy based on particle filter is proposed to estimate the state.By iterative learning,the state estimate of previous batches is employed as the measurements information for the current batches,and the data information of previous batches is fused to improve the estimation accuracy of current batches.Compared with the previous batches,the initial state error decreases gradually and tends to converge as the number of batch increases,and the estimation effect of the whole batch is improved.The effectiveness of the proposed state estimation algorithm is indicated by its application in the digital simulation and beer fermentation process.