METHOD OF DETECTING ABNORMAL CHANGES IN THE FUNCTIONING OF A TECHNICAL OBJECT USING SINGULAR VALUES DECOMPOSITION OF EXPERIMENTAL DATA MATRIX

The behavior of the dynamic system is characterized by a large set of experimental data. The character of this data changed in abnormal case of functioning or in system faults. The identification of the considerable changes from measurement results almost no using statistical methods is the feature of the article. For this purpose,the experimental data matrix and the singular value decomposition (SVD).The change in singular values occurs as a result of a significant change in the nature of the data.To make a decision regarding the system transition into anomalous mode, the distance in Euclidean space between the set of singular values for normal conditions of system operation and the set of such values for possible anomalous conditions is estimated.If this distance will exceed the given threshold, then a decision is made to transfer the system to abnormal mode. Threshold determination can be carried out using statistical methods.