Method of machine technical state assessment on the basis of joint signal analysis

Abstract

Joint analysis of two signals which are observed during machine operation allows us to acquire information on such phenomena occurring within the machine which cannot be obtained as the results of an analysis of single signals. This knowledge makes it purposeful to assume that the joint analysis of a major number of signals allows us also to acquire new information on the machine state. In this paper the concept of joint signal analysis based on a signal analysis in both “micro” and “macro” time domains was described. In order to calculate values of joint features of signals and determine the machine state on the basis of the joint analysis results the generalized data window was introduced. In order to limit a number of simultaneously considered signals and determine a relation between signal feature values and machine states, the procedure based on an application of additional virtual signals was proposed. In order to verify the elaborated method an active diagnostic experiment was carried out. The experiment result was a set of learning data, which enables us to construct a diagnostic classifier. Exemplary results of classification of technical state of a machine were also presented.