On-line Detection For CVT Based On Multi-parameter

Capacitor Voltage Transformer(CVT) is an important equipment providing signals for power system protection and measurement, its performance is closely related to the economic and safe operation of power system. Domestic and foreign experts have done a lot of research on state detection for CVT, although power-off maintenance can prevent faults, it need interrupt power supply and influence operation of power system. There are two main uninterruptible power detection methods, live detection and on-line monitoring. The existing live detection methods, such as infrared detectors, etc, can not quantitatively analyzed CVT capacitance value, the dielectric loss and some other parameters, so there are still defects for preventing faults. Now, there are not so many on-line monitoring methods for CVT, and the monitoring signals are bus voltage and leakage current. So the monitoring signals are not enough, only can be used for testing total capacitance and dielectric loss of value, can not reflect the operating status of each divider capacitors and intermediate transformers unit.As live detection can not realize quantitative analysis of CVT parameters and signals for on-line monitoring are less, a live detection method with multi-parameter was put forward, which was very flexible. This method can realize live detection of CVT with detection device.To solve the problem that detection signals are less and can not reflect the state of CVT operation effectively, a lot of work was done, such as reviewing literature and survey. The four signals, consisting of bus voltage, secondary voltage, the current flowing through medium voltage capacitor and current flowing through electromagnetic unit were select as detection signals to calculate the ratio of two voltage signals and their phase angle difference, the ratio of two current signals, coefficient of variation in total capacitance, high voltage capacitance and intermediate voltage capacitance and use these six state parameters to reflect the CVT operating state.In order to study the relationship between six state parameters and CVT operational failures, a 220 kV CVT equivalent simulation model was established with ATP / EMTP simulation software. In order to verify the correctness of the equivalent simulation models and simulation conclusions, an experiment was done. In addition, a BP neural network algorithm was established to detect the state of CVT based on relationship between six state parameters and CVT operational failures.A CVT live detection device was developed with hardware and VC++ program, which can realize the Collection, storage and computation of four signals. In order to eliminate the error caused by that signal sampling was not at the same time, a standard frequency signal was used as a reference signal. Through calculating the phase angle difference, the delay time can be obtained, which greatly inhibited the sampling error. In order to inhibit the error due to the spectral leakage and picket fence effect, periodic phase shift method was used to calculate periodic phase shift method to calculate the frequency of the sampled signals, and interpolation function was used to recover these four signals to achieve the goal of integrated period sampling, greatly inhibiting the error spectrum leakage and fence effect caused.In order to test the reliability of the live detection method and the detection device, an experiment was conducted on four CVT devices in State Grid Ganzhou Power Supply Company, and the experiment result was same to maintenance records, proving the detection method and apparatus useful for CVT live detection.