【摘要】 提出一种随机非线性系统状态和参数同时估计的神经网络新方法，并证明了该方法的无偏性和最小方差性。将其用于乙醇间歇发酵器的状态和参数估计，结果表明估计值与实验值相吻合。此方法对噪声特征无特殊要求，对初始状态估值不敏感，对初始参数估值具有一定的鲁棒性，可利用有限的状态量测信息在线估计不可测量的状态变量和物理参数。更多还原

【Abstract】 On-line state estimation for a kind of bioreactors is developed using presently available off-line measurements via functional -link neural network. The performance of nonbias and minimal error variances of this on-line state estimator are proven. It is applied for the states and parameters estimation of ethanolfed--batch fermentation. The results of simulated and experimental on-line state estimation show goodagreement. It has not special requirments for system noise and measurement noise distributions, is quite insensitive to its initial state estimation, and has strong robustness on the estimated initial parameter values.更多还原