Time-Domain Characteristics And Calculation Model Of Corona-generated Audible Noise On DC Conductors

With the rapid construction and operation of the ultra-high voltage direct current (UHVDC) transmission lines, the corona-generated audible noise has become one of decisive factors in the selection of conductor and configuration design. It is of significance to investigate the characteristics of the audible noise, its correlation with the corona discharge and prediction method for giving an instruction for the design and operation of the HVDC transmission lines. This thesis adopts the time-domain measurement method and de-noising method to investigate the time-domain characteristics of the DC corona-generated audible noise. The statistical characteristics in time-domain of audible noise from single corona source, two corona sources and conductors are obtained. The correlation between the audible noise and corona current in time-domain is presented. Furthermore, the stochastic models in time-domain for the sound source from single corona source and the stochastic prediction method for the audible noise from single conductor are proposed. The validity of the model and prediction method is given by comparisons of the measured and calculated sound pressure levels at the receiving point. The work of the dissertation may promote the acknowledgement of characteristics and generation mechanism of the DC corona-generated audible noise, providing a reference for the prediction analysis of audible noise from HVDC transmission lines.The time-domain measurement system for corona-generated audible noise is established and the time-domain waveform of audible noise from single corona source is obtained. Compared with the traditional frequency-domain measurement method through which only the amplitude-frequency characteristics can be obtained, the time-domain measurement method can obtain real-time sound pressure and may be more effective to reveal the intrinsic characteristics of corona-generated audible noise. Besides, a de-noising method based on the measured time-domain waveform is proposed for removing the influence of background noise in the corona-generated audible noise from single corona source. Through the de-noising method, sound pressure pulses from corona discharge can be extracted, and the real-time background and reflection sound from ground can be removed, which may lay a foundation for the accurate analysis of audible noise. Furthermore, on the basis of the principle of sound level meter and the random nature of corona-generated audible noise, the method for transferring the time-domain waveform of audible noise to sound pressure level is proposed.A typical waveform for characterizing the sound pressure pulse is proposed and the validity of the typical waveform is given by comparing the time-domain and frequency-domain results of typical waveform and measured waveform. The statistical characteristics of the sound pressure pulses are analyzed. Besides, the correlation between audible noise and corona current in time-domain is obtained through the measured waveforms of audible noise and corona current, which may provide a new method to analyze the time-domain characteristics of the audible noise.The propagation model for audible noise from single corona source is proposed. On the basis of measured waveforms of audible noise at the receiving point and the propagation model, the waveform of the sound source is inverted and statistical characteristics of sound pressure pulses of sound source are obtained. Furthermore, a time-domain stochastic model for simulating the sound pressure pulses from single corona source is proposed and construction of the waveform of sound source is achieved. The validity of the proposed stochastic model is testified by measured and calculated sound pressure levels at the receiving point.On the basis of the one-to-one relationship of the sound pressure pulse and corona current pulse, the audible noise waveforms from two corona sources are separated. The characteristics of the sound pressure and sound pressure level from two corona sources are analyzed. Besides, statistical characteristics of the sound pressure pulses from single conductor are obtained through the experiments carried out in the laboratory. A stochastic prediction method in time-domain for simulating the audible noise from conductor is proposed. The validity of the proposed method is given by comparising the measured and calculated sound pressure levels at the receiving point.Furthermore, the relationships between the sound pressure level and pulse parameters, conductor parameters are obtained through the proposed stochastic prediction method. The work presented in this thesis can lay a foundation for prediction of the audible noise from HVDC transmission lines.