| Rotating electrical machines are widely used in every industrial department. With industry process being more and more automatic, flexible, integrative, intelligent, exact, and the unit capacity of rotating machines (RM) being more and bigger, or the effect of RM being more and more important, the economic loss caused by fault or unscheduled maintenance of RM becomes very huge. All these show that the condition monitoring and fault diagnosis of RM is imperative and necessary, which also brings a large amount of economic benefit and social benefit. Under such a background, this thesis systematically summarizes the contemporary monitoring and diagnosing technologies of PM, finds the shortages of the previous methods firstly. New inductances calculating way based on winding function, and three different fault monitoring and diagnosis methods used in squirrel cage induction motors are proposed then. Stator winding temperature models for large turbo-generators and the diagnosis way for the hollow strand blockage are proposed at the same time. Finally a new approach to detect commuting performance in DC motors is proposed.There are three main faults in the squirrel cage induction machines: coils short circuit in stator windings, broken rotor bars and rotor eccentricity. Although characteristic of faults is already known, there is no complete theoretical analysis yet. With the views of basic electromagnetic and electrical machine theory, this research gives interpretation of mentioned three faults. The simulation of AC motors is important to the design, control, operation and fault research of motors. Among the parameters in the simulating models, the inductance of stator and rotor are most important. The existing inductance computational formulae are suited to the normal condition. These formulae cannot be used to fault condition (especially eccentricity faults). According to the winding function and the expression of air gap, this thesis gives out computational formulae for the stator and rotor inductance parameters in the induction motors under both normal and fault conditions. Simulation results tally with that of experiments, which indicates correctness of the models and the parameters computation. The previous methods to analysis the spectrum of line current always cannot highlight the characters of the broken bar since the frequency of the fault component is very close to the power frequency. Therefore this research proposes three different fault diagnose method: (i) The method based on Hilbert mode spectrum analysis: it just uses the sample of single-phase current, and spends less hardware, which is mainly used in the rotor fault detection; (ii) The method based on vector control: it can be used to diagnose all kinds of faults in stator and rotor, needs sample each pahse's current and voltage, as well as the parameters of the stator and rotator. The merit of this method is that it is easy to integrate with the control system; (iii) The method based on spectrum analysis of instantaneous power: it can also detect all kind of stator and rotor faults, where there is not necessary to get parameters of the resistance and inductance of the motor. The three methods mentioned above have been proven with inference processes, and the theoretical analyzing conclusions tally with the results of experiments. All these indicate that the new methods are effective in their own application domains respectively. The hollow strand blockage in the stator winding bars is one of the most common faults in large turbo-generator, which is of destruction to the machines. In power factories, staffs mainly detect faults by means of temperature warning. However, before the sign of warning, the winding have already been destroyed and hurt. Within this way, information reflected by the temperature measure sensor is not sufficiency used in the fault detection. This research proposes two temperature models: one for the stator Teflon outer water, another for the site of the temperature measuring sensors at the slot. These two...
|
PDF Full Text Request