渐消记忆递推最小二乘法在呼吸力学参数估计中的应用

油酸及组胺两种急性肺损伤犬模型检验了渐消记忆递推最小二乘算法辨识跟踪慢时变呼吸力学参数的效果,除肺力学参数外,还对胸-肺系统总的呼吸力学参数进行了估计。此外,还与用最小二乘曲线拟合法得出的参数估计结果作了比较。结果表明:渐消记忆递推最小二乘(RLS)法精度高、收敛速度快,适于跟踪慢时变的呼吸力学参数。作者认为,针对临床监测目的,以选用一室一阶线性呼吸力学模型跟踪呼气相的呼吸力学参数为宜。

Application of Recursive Least Square Algorithm in the Evaluation of Respiratory Mechanical Parameters

The monitoring of respiratory mechanics of lung injury patients isvery important in diagnosis or monitoring of disease,appreciation of therapy and evalua-tion of prognosis.We examined the effect of recursive least square algorithm(RLS)intracing the slow time-varying respiratory parameters on canine acute lung injury modelsinduced by oleic acid or histamine.Besides lung mechanics,the total respiratory parame-ters of chest-lung system were estimated at the same time.This result was compared withthat of curve fitting(LS)method also.The results obtained with LS method was consis-tent with that obtained with RLS when estimating compliance parameter,but R parameterobtained with LS method was not satisfactory.It suggested that RLS method is highly pre-cise,fast convergent and suitable for tracing slow time-varying respiratory parameters.The RLS method can be used for on-line identification.As for clinic monitoring,wethink it is better to use onecompartment linear model in tracing respiratory mechanics ofexpiratory phase.