Translated Abstract
The inter-turn insulation of inverter-fed machine suffers the repetitive steep-fronted impulse voltages, which gets vulnerable and shows significant reduction of life-time with the existence of insulation defects. It is very difficult to diagnose the inter-turn insulation defects by traditional methods and techniques. In this paper, the achievements of research on detection, recognition and locating of insulation defects for machine winding based on 300kW/2kV traction motor winding is presented. The research on inter-turn impulse voltage induction method, development of novel near-field partial discharge (PD) sensor, simulated tests with artificial defects, feature extraction and diagnosis of detected signals have been carried out. It is beneficial on the prevention of insulation defects in manufacture, and timely maintenance during operation, which is of great importance for raising quality and reliability of machine insulation system.
The distribution of impulse voltage along a winding is uneven and the inter-turn failure risk exists not only in coils near winding terminal, both indicate the inherent limitation of traditional surge test. Due to the traditional electric diagnosis on the stator winding terminals of electrical machine are not effective enough to detecting, identifying and locating the defects in the inter-turn insulation, especially for the medium or small-sized machines, an impulse generator with removable external impulse current excitation unit used for inducing the impulse voltage on the testing stator coil have been developed. The impulse induction test system is modeled and simulated with PSPICE on the basis of experimental results, and the relationships between the induced voltage waveform parameters and the test system circuit parameters have been obtained. A computer-aid test configuration design program is developed based on a 3rd order equivalent circuit model solution algorithm. Four basic evaluating parameters are proposed to evaluate the effectiveness of the testing with reasonable range. Many experiments have proved that induction units with opposite polarity in series applied at adjacent target coils is a simple and effective option.
The response signal of inter-turn defect under impulse voltage has local transmission and high-frequency attention characteristic along the winding. The technically feasible option to achieve fully capture of PD signals is near-field measurement. A novel non-contact electromagnetic induction sensor suitable for inter-turn PD detection under induced impulse voltage is designed. Taking advantage of 1K107B ferrum-based nanocrystalline alloy thin strip applied in the magnetic core, the sensor is sensitive to the high frequency low amplitude PD current signal and have flexible and thin framework on consideration of the spacing limitations and convenience of winding-end measurement. The sensor suitably applied in near-field measurement straddling on the straight part of winding-end has high sensitivity and good convenience to locate defects. Its transfer characteristic within 0~100MHz is studied by theoretical analysis and electromagnetic finite element simulation. The amplitude is bell-shaped, the maximum gain can reach 104V/A near 55MHz, and the phase curve is waterfall-shaped. The influence of the number of turns of the coil, the thickness of the core and the diameter of the coil on the transfer characteristics of the sensor is studied to meet the needs of different research objects and testing requirements.
Artificial void insulation defects with 200~2000V impulse partial discharge inception voltage (IPDIV) and conductive defects with 0.1~700Ω resistance are prepared to simulate actual inter-turn defects, in order to verify the feasibility and effectiveness of the insulation defect detection method under induced impulse voltage. Inter-turn defect diagnosis can be achieved by summarizing the time-frequency features from massive test data. Research shows that, the wavelet energy distribution and wavelet packet singulars can recognize the existence of conductive and void defects, respectively. The amplitude and time-lag of inter-turn PD pulses are usually Gamma distributed, whose statistical features indicate the detail information of void defect. The severity of the void defect can be diagnosed by K-S test.
The important goal of inter-turn defect detection is to ensure the inter-turn insulation is in a healthy state, so as to make reasonable maintenance plan and avoid potential failure risk. An electrical-thermal aging test system is established. Two kinds of aging programs with twisted-pair and module coil specimen of inter-turn insulation are made according to real stress condition of 300kW/2kV traction motor. The test results indicate that, the inter-turn void defect grows with PD, turns into conductive defect after local carbonization, and finally leads to inter-turn fault. The mark of the critical point is that, the maximum response pulse obtained from the sensor reaches 50V under 500V inter-turn impulse voltage. The information detected under induced impulse voltage can reflect the aging process, which can be regarded as an effective parameter to evaluate the aging state of inter-turn insulation.
Translated Keyword
[Defect diagnosis, Electrical-thermal aging, Inter-turn insulation, Inverter-fed machine, PD sensor]
Corresponding authors email
