ОЦІНКА ОСОБЛИВОСТЕЙ БАГАТОВИМІРНИХ ВИПАДКОВИХ ПРОЦЕСІВ НА ОСНОВІ ЇХНІХ ВІЗУАЛЬНИХ ОБРАЗІВ

Abstract

Despite the wide implementation of information technologies in diagnostic systems, visual control of the processes that characterize the object state remains relevant, especially when the number of such processes is large. An analysis of the possi-bilities of visual control for multidimensional random processes, the realizations of which are processed in measurement infor-mation systems on complex technical objects, was carried out. For this, these processes are transformed into a visual image, that represented a set of curves on a plane, which are obtained for each discrete moment of time. Any curve is the sum of products of the realizations values for each random process at a specified discrete moment of time on the corresponding or-thogonal function that depends on some generalized argument. Lagrangian functions are used in the article.The purpose of the article is to check the effectiveness of a new method of visualizing realizations of a multidimensional random process for extreme cases and to develop proposals for detecting the parameters jumps of one or more processes.For the first time, a method of visual detection of the parameters jumps of realizations for one or more random processes is proposed. The essence of the method is based on the subtraction of a visual image, that is, a set of functions in the absence of a jump of one or more process parameters from a similar image that concludes jumps of parameters. In practice, the first image should be a statistical average for the conditions of normal functioning of the technical object. After subtraction, a new visual image is created, which shows anomalies caused by jumps in process parameters. The usefulness of the method is demonstrated on the example of processing multidimensional experimental data, which were matched with generalized visual images. The functioning of this method is verified also for extreme cases of processing white noise and a deterministic process. The obtained visual images also reveal the features of the behavior of multidimensional random processes.