Two-dimensional trend analysis of time series of complex technical objects diagnostic parameters

An approach to assessing the relationship and differences of multidimensional trends is proposed and justified. The approach is based on construction of multidimensional arrays from time series of registration data of the diagnosed objects technical parameters. To identify the similarities and differences of time series trends, a pairwise joining of their samples with the same arguments is proposed. The unified time series has the counts in the form of complex numbers and is analysed by the proposed improved principal components method. The proposed method permits dividing the parameters of the object condition into groups that have trends of the same type, which allows localising faults and increasing the reliability of diagnostic conclusions about the technical condition of the object. The a priori statistical model of data generation adopted in the studies was chosen as a model of deviations of the diagnosed objects parameters from the nominal values.An approach to assessing the relationship and differences of multidimensional trends is proposed and justified. The approach is based on construction of multidimensional arrays from time series of registration data of the diagnosed objects technical parameters. To identify the similarities and differences of time series trends, a pairwise joining of their samples with the same arguments is proposed. The unified time series has the counts in the form of complex numbers and is analysed by the proposed improved principal components method. The proposed method permits dividing the parameters of the object condition into groups that have trends of the same type, which allows localising faults and increasing the reliability of diagnostic conclusions about the technical condition of the object. The a priori statistical model of data generation adopted in the studies was chosen as a model of deviations of the diagnosed objects parameters from the nominal values.