Investigation Of Identification Algorithms In Respiratory Parameter Tracking

The respiratory system is one of the most important system in humanbody, is an important part in the body for gas exchange with the outside world. In clinical practice,monitoring the status of the patient in realtime,accurately noninvasive estimate the time-varying parameters in mechanics respiratory, and frequent corrective adjustment the ventilator,is one of the important means to help the patients obtain optimal respiratory assistance. The research in this paper, is to noninvasive estimation respiratory time-varying parameter by identification algorithm, which is rather applicable and universal,to clinical medicine and improvement the technology of breathing machine is great significance.Respiratory system model is the basis for the respiratory parameters to estimated, the nonlinear time-varying model is difficult for estimate the parameters online,and the model is too complex for algorithm tracking and analysis. Therefore, this article chooses the first-order linear model is the most widely used, mainly related to the following practical need:(1) in clinical monitoring, online identification for the purpose of work requirement model is simplified; (2)Can diction inpathology, artificial ventilation conditions stable results, direct physiological significance given information.This article is based on recursive least square method through calculating the minimal sum of squared residuals of the best parameter estimates for airway resistance, elastic resistance and pleural cavity pressure,and forgetting factor by setting the index to gradually reduce the weight of past data and increase the weight of the current data, in order to reach the purpose of online parameters estimation, and solved the "data saturation" problem. Considering the different respiratory parameters,such as airway resistance, elastic resistance and pleural cavity pressurel, the change rate of each different, the improved identification algorithm is proposed by tuning matrix,by set up tuning factors for different rate of respiratory mechanics parameters, independent tracking to meet the different parameters, so as to solve the parameter estimation between data related to rate of change of different parameters.Simulation results show that recursive least squares of single tuning factor estimate the parameters of the basic deviates from the true value,estimate of the elastic resistance is affected by airway resistance step change, especially for pleural cavity pressure, basic unable to capture any reference information. And improved identification algorithm increased tuning matrix, make up for the deficiency of the single tuning factor, effectively strengthen the estimate of the time varying parameters, improves the accuracy of parameter estimation, algorithm haves good noise resistance and strong robustness.