论文部分内容阅读
监护信息系统中通常采用系统建模的方法对监护数据进行分析处理、并识别异常状况。通常这些模型是在模型结构确定的条件下,应用监护数据辨识获得。实际检测获得的监护数据通常包含大量异常值,这会严重降低模型辨识的准确性。提出了一种监护信息系统中异常值分析处理方法。应用Hampel辨识器算法,识别监护数据中异常值出现的位置;并采用kalman滤波器算法的方法对于出现异常值的数据点进行数据重构,实现监护信息系统中出现异常值分析处理。通过应用PhysioNet生物医学信号研究资源中的两组数据集,包括心率和中心静脉压,进行实验研究,结果表明此方法对监护数据异常值分析和处理中取得很好的效果。
Guardianship information system usually adopts the method of system modeling to analyze the guardianship data and identify the abnormal situation. Usually these models are obtained under the condition of the model structure identification and identification by the monitoring data. Custody data obtained from actual tests often contain a large number of outliers, which can severely reduce the accuracy of model identification. A method of analyzing and processing outliers in monitoring information system is proposed. The Hampel recognizer algorithm is used to identify the location of outliers in the monitoring data. The kalman filter algorithm is used to reconstruct the data points with outliers, so as to realize the outlier analysis and processing in the monitoring information system. Two sets of datasets from PhysioNet Biomedical Signal Research Resources, including heart rate and central venous pressure, were used to perform the experimental study. The results showed that this method achieved good results in the analysis and treatment of outliers of monitoring data.