Monitoring voltage harmonics represents one of the most important tasks in power quality assessment. In particular, the employed instrument transformer plays a key role in the achieved accuracy. Its harmonic measurement performance is typically evaluated by measuring its frequency response function. However, nonlinearities may have a non-negligible impact on measurement uncertainty: for example, this occurs as far as inductive voltage transformers are considered. This paper proposes a simple technique allowing the compensation of the most significant nonlinear effect, which is the harmonic distortion produced by the large fundamental primary voltage. The method is firstly derived and introduced by means of numerical simulations, and then implemented through a proper experimental setup. Results highlight remarkable accuracy improvements when realistic voltage waveforms are measured.

The most recent IEC standards about voltage transformers warn about nonlinearity, which may have significantly impact on the harmonic measurement performance. However, there is a lack of scientific literature about this topic: usually their characterization consists of frequency response measurements, which are clearly not able to capture the nonlinear behavior.In this paper, an innovative approach based on simplified frequency domain polynomial models developed by the authors is proposed. The method is applied to two different medium voltage inductive transformers. Models are identified and validated with a large set of realistic primary waveforms injected by a proper setup. Experimental results confirm the remarkable accuracy of the proposed models especially for low-order harmonics, which are the most affected by nonlinearity.

Voltage instrument transformers are usually tested at the rated frequency. In order to assess their performance in measuring harmonic components, typically, the frequency response function (FRF) is evaluated. Therefore, this conventional characterization does not consider nonlinear effects that may have a nonnegligible impact on the accuracy, especially when the transducer under test is represented by an inductive voltage transformer (VT). In this paper, a simple procedure for the characterization of voltage instrument transformers is presented. The method is based on the concept of best linear approximation of a nonlinear system. It requires applying a class of excitation signals that resembles the typical voltage waveforms found in power systems. Results consist of the FRF that permits the best linear compensation of the transducer response, and sample variances that allow quantifying the impact of noise and nonlinearities on the accuracy. The method is presented and explained by means of numerical simulations. After that, it has been applied to the characterization of a conventional inductive VT. Experimental results show how the accuracy of the transducer under test is heavily degraded by nonlinear phenomena when low-order voltage harmonics are considered.

No abstract

The Volterra approach to the modeling of nonlinear systems has been employed for a long time thanks to its conceptual simplicity and flexibility. Its main drawback lies in the number of coefficients, which rapidly grows with memory length and nonlinearity order. In some important cases, such as power system applications, the input signal is periodic and contains a fundamental component that is much larger with respect to the others. This peculiarity can be exploited in order to dramatically reduce the number of coefficients defining the frequency-domain Volterra model with slight drawbacks in terms of accuracy. A systematic procedure for the definition of simplified, frequency-domain models of arbitrary order is proposed. Thanks to the simplification, very high orders of nonlinearity can be managed. The proposed approach has been employed to model the behavior of two electrical devices with different amount of nonlinearity, and that of a power grid containing linear and nonlinear loads. Accuracy is discussed and compared with that obtained with a conventional Volterra model defined by a similar number of coefficients. Results show the effectiveness of the approach, which is particularly suitable to model and test voltage and current transducers as well as other ac power system devices