Understanding Changes in Global Behavior Due to Control Location 87 where the PDF is centered at a mean value µi with distribution width σi, and an error metric value of xi. We will be using the dB error described in Equation 2 as the xi value. The mean value of the PDF is set to zero, indicating perfect agreement between the response and specification, and the distribution is set such that 3σ on the PDF is equivalent to 6 dB, which is a common abort limit in environmental testing. The PSDPM value is then normalized by a PDF with a mean and error metric value of 0 to ensure the final value will be between 0 and 1: PSDPM= 1 nfreq nfreqX i=1 g(edBi, 0,σ 2 i ) g(0, 0,σ2 i ) (4) where we will be computing the PSDPM at each frequency line then using the RMS value for assessment. The PSDPM offers a different perspective into the error between field and environmental response. At each frequency line, the PSDPM gives the probability that the dB error between the field and environmental response is zero, that is that the responses are the same. For comparison to other scalar-value metrics in this paper, we compute the RMS of the PSDPM across frequency lines. We consider this PSDPM RMS as an alternative measure of how well the specification/field environment was achieved by a given test. Difference in response Before we can make an assessment of how similar the global response of the GOBLET was between the lab and the field, we must first ensure that the control DOFs were controlled well in the environmental test. To assess this, we will first use visual inspection, then apply the error metrics from the previous section. In Figure 4, we see agreement when controlling to Fig. 4 Response at controlled DOFs compared to reference specifications derived from field data with±3 dB alarm limits and±6 dB abort limits.
RkJQdWJsaXNoZXIy MTMzNzEzMQ==