11.16 – information-measuring systems. – Institute of Electrodynamics of Ukrainian National Academy of Sciences, Kyiv, 2001.
The dissertation is devoted to the problems of statistical approach to the technical diagnosing of complex electrotechnical and power equipment, which works in a cyclical mode. The increasing of reliability of such equipment, in particular of diesel generators, is very important problem. It can be solved by development of specialized diagnosing and monitoring systems. Most of existing methods and tools of technical diagnosing of diesel generator use deterministic methods. So, they do not take into account the stochastic nature of diagnostic signals and processes, which accompany with an operation of equipment. Therefore the new diagnosing systems, based on the new methods and mathematical models, should be developed.
Functional diagnosing of equipment implies the measurement of different parameters or signals generated by object while it operates. For diagnosing of diesel generator pressure and temperature of gases in the cylinders and vibrations of units are used most frequently. In the dissertation it is offered to use for diagnosing at the same time the information, which is supplied by vibrations of a diesel generator and deviation of an angular velocity of its crankshaft. The simultaneous analysis of these two signals expands diagnostic space and the reliability of diagnostic solutions grows.
The success of diagnostics appreciably depends on mathematical model, on which the diagnostics method is based. In the dissertation the new models of vibrations and deviation of an angular velocity of crankshaft were elaborated. These models are linear periodic random processes. The main feature of these models is that they take into account both cyclical and random nature of physical processes, which are used as informational for diagnosing. It is known that for linear process the canonical representation of characteristic function exists. In this representation the kernel function and parameters of generic process are used. It gives a possibility to express all probabilistic parameters of linear process via this parameters and to carry out whole probabilistic analysis of such processes. It is used for evaluation the accuracy of information signal measurements.
Vibrations, caused by each cylinder of diesel generator, propagate through the casing to the measurement point. The casing is considered as linear system with many resonance frequencies. The pressures in the cylinders are the input influences of this system. Since the influence is random periodic process, the vibrations of diesel generator will be linear random periodic process. Model of the deviation of crankshaft angular velocity is based on the assumption that crankshaft is solid body. The crank mechanism of diesel generator transforms the pressure in each cylinder into the turning moment. The change of this moment causes change of angular velocity. Therefore we can consider the deviation of angular velocity as linear process with periodic kernel function and periodic generic process. The kernel functions and parameters of generic processes of above models can be experimentally determined on the diesel generator.
Method of measured signals processing is very important also. To carry out the diagnosing of diesel generator the spectral analysis is used in the dissertation. Main feature in this case is that information signals are stochastic processes and are non- stationary. The spectral analysis of some class of non-stationary processes, named harmonizable, was developed by Loeve. But the equations were obtained using complex harmonics. So, we have negative frequencies, not existing in nature. The new formulas for spectral analysis of real harmonizable process were derived in the dissertation. They give the possibility to carry out the spectral analysis using real functions and non-negative frequencies only. Based on the obtained results it was ground, that the jumps of two-dimensional spectral functions of proposed above informational signals can be used as a diagnostic parameters for determining the status of diesel generator.