*4.2. Data Evaluation*

In the following, the evaluation of the data acquired using the oscilloscope is discussed, to make the presentation of the results more comprehensible. The recorded sequences with 20,000 data points each were evaluated individually in MATLAB® to determine the impulse current amplitude ˆ*I*PD and the time difference between subsequent sequences Δ*t* (Figure 7). In this investigation, the time differences between two sequences were a measure for the time differences between subsequent impulses, since the length of one sequence of 1 μs was in most cases negligible with respect to the time difference Δ*t* >> 1 μs. If one sequence contained more than one partial discharge impulse, this would be evaluated separately.

**Figure 7.** Sample sequences for the data evaluation (negative protrusion, *p*SF6 = 0.1 MPa, *U* = 2 · *U*i).

To achieve a meaningful data evaluation, the following definitions were made and applied to every recorded sequence: A noise level was calculated with a moving average out of 1500 data points multiplied by a factor of five. A PD current impulse needed to be larger than this calculated noise level and higher than a defined minimum impulse amplitude ˆ *I*PD min to be evaluated as a PD event (Table 2).


**Table 2.** Parameters for the data evaluation.

\* No measurements possible because the measurement voltage was too close to breakdown voltage.

The parameter ˆ *I*PD min was determined depending on the measured impulse current amplitudes, since the noise level was dependent on the chosen measurement range of the oscilloscope. If one sequence contained more than one single impulse, the definition of a minimum time difference between subsequent impulses in one sequence <sup>Δ</sup>*t*imp min was necessary, to avoid the oscillations following an impulse being evaluated as another PD impulse.
