Post-measurement processing of electric power quality data

This paper addresses the issues of post-measurement processing of data collected in a power quality assessment study. Three broad types of post-measurement processing objectives are considered: to enhance accuracy; to estimate data; and to reduce the volume of the collected data. The methods used to enhance accuracy are bad data identification and rejection. Averaging is discussed as a method to "filter" measurement error. The methods used for data estimation are state estimation techniques in both the lime and frequency domain. The methods used to reduce the volume of the collected data are based on the calculation of marginal and conditional probabilities and expectations. The integrated use of these techniques in an instrumentation system for power quality assessment is discussed. Table 1 shows a comparison of post-measurement techniques for power quality data assessment. The main results of the paper are applied to harmonic measurements in power systems. Bad data rejection, based on out-of-range statistical methods is described. Accuracy enhancement using a frequency domain technique based on frequency response correction is described. This technique is equivalent to time domain convolution and this technique is illustrated. A second broad post-processing method for processing power quality data relates to the minimization of measurement noise. State estimation is used for this task. The third method illustrated is intended to reduce the volume of collected data. Statistical techniques are used to calculate conditional probabilities and to obtain probability indices within given, prescribed accuracy limits.