A novel method for predicting power system voltage harmonics with an artificial neural network is presented. The method is based on the backpropagation learning technique for feedforward neural networks. The promise of the proposed method in harmonics prediction is shown. In order to demonstrate its effectiveness, the proposed method is applied to voltage harmonics observed through a personal computer based measurement system and the performance is compared with that of conventional methods. >

An artificial neural-net based method for predicting power system voltage harmonics

Findings from a study on the harmonics generated by a 15 MVA three-phase 13.2 kV 60 Hz to single-phase 24/12 kV 25 Hz static frequency converter (SFC) or cycloconverter are summarized. At light loads, some of the characteristic harmonics (e.g. 10 Hz, 110 Hz) have magnitudes that exceed the fundamental frequency. Since the three-phase components at these frequencies do not necessarily nullify each other, many frequency components, in addition to zero-sequence components, can be measured in the neutral (residual) current. Because of low system impedance at harmonic frequencies, the voltage at the 13.2 kV substation supplying the SFC is not severely affected. A worst-case scenario for the voltage spectrum is presented. The corresponding distortion factor is calculated to be 4.12%. The power factor of an assumed parallel load is calculated using this scenario, and the effect of the SFC harmonics on the power factor is found to be insignificant. However, depending on the system impedances, the shunt capacitors for power factor compensation at the assumed load may be ineffective due to harmonic resonance. >

Impact of cycloconverter harmonics

This paper deals with the correction of the power factor of the load when the bus voltage contains harmonics. The various optimization criteria are discussed for determining the optimal capacitor for power factor correction, The performance of the obtained capacitor is discussed by means of numerical examples.

Power Factor Correction for Distorted Bus Voltages

A method is presented for finding the optimum fixed LC combination for power factor compensation of nonlinear loads where both source and load impedances are represented as randomly time-varying quantities. The distribution characteristics of these random variables at a supply bus can be experimentally determined by using a microcomputer-based parameter estimator. If maximum power factor is the main concern, the optimal LC combination corresponding to maximum average power factor is obtained for any given compensator cost. Laboratory tests show the optimization algorithm and the parameter estimator to be very efficient and accurate. >

Optimal LC compensators for nonlinear loads with uncertain nonsinusoidal source and load characteristics

Comparing capacitive and LC compensators for power factor correction and voltage harmonic reduction

The goal of this paper is to determine the effect of time variation of system impedances and voltage harmonics on the value of optimum capacitor for power factor correction at busses with nonsinusoidal voltage. Two types of 24 hours time-variation of voltage harmonices and Thevenin impedance are assumed. The equivalent load impedance is also considered time variable and assumed to contain a large proportion of induction motors. The daily energy losses are computed and graphed in function of the shunt capacitance used for power factor correction. The results of this study indicate that in order to avoid resonances and to find the optimum capacitor the time variation of harmonics and system impedances must be known as precise as possible.

Roham T. Saleh

Papers

Optimum Shunt Capacitor for Power Factor Correction at Busses with Lightly Distorted Voltage