Fault detection and isolation of ECG process using kernel principal components

In this paper, a study of the regularity of the electrocardiogram is given by the integration of a process diagnostic tool that is kernel principal components. It is a question of detecting the defects which exist in the set of parameters of the process of the electrical activity of the cardiovascular system: segments of the ECG and amplitude of the waves. The kernel ACP which is a powerful tool in fault detection is applied to a multivariable matrix of data from the ECG process. Exceeding the control thresholds of the SPE statistic proves the existence of a defect in the data studied. The calculation of contributions is introduced for the isolation of defects.