| This dissertation deals with applications of the wavelet transform and other digital signal processing techniques to solve various problems associated with (1) transient power line disturbances and (2) wireless communication systems.; With regard to power line disturbance analysis, first, a real time power disturbance detection method is developed which consists of an adaptive prediction error filter to extract the disturbance from the power line signals, and a SG-CA CFAR (stop-and-go cell-average constant alarm rate) detector to provide robust disturbance detection. In our tests this algorithm achieves a 100% detection rate and an exact disturbance starting time for distribution and transmission line disturbances. Second, when the detected disturbance signal is transmitted and archived at the monitoring station, we need to compress the disturbance signal because of network traffic bottlenecks and the need to efficiently archive the data. Thus, we develop a compression scheme with a variable and high compression ratio using EZWT (embedded zerotree wavelet transform). This approach provides a high compression ratio ∼15:1 with low NMSE ∼10−4 for distribution and ∼10 −1 for transmission disturbances. The Daubechies 6 wavelet is recommended for distribution and transmission line disturbance compression based on testing of actual power line disturbance events. Third, as the first step of power quality assessment, the disturbance events should be classified into various disturbance categories. However, too many events are recorded to be classified by humans. Thus, we develop a classification method using a rule-based classifier for the time-characterized disturbances, e.g., sag and interruption, and wavelet transform HMM (hidden Markov model) for the frequency and time dependent disturbances, e.g., capacitor switching, normal variation, and impulses. The correct classification ratio for 670 actual disturbances is about 98.7% without rejecting any events. Fourth, as a final step of instantaneous power quality assessment, we develop an IDE (instantaneous disturbance energy) index which extracts only disturbance energy using the Teager energy operator. This approach is computationally simpler than the previous wavelet approach.; As to communication systems, the work falls in two categories. The first one deals with the reduction of PMEPR (peak-to-mean envelope power ratio) in multicarrier CDMA (code division multiple access) systems. This algorithm utilizes the discrete inverse wavelet transform for the modulation part and discrete wavelet transform at demodulation part. The resulting PMEPR is 8.5% lower than the fast Fourier transform based method. The second application is focused on nonstationary jammer excision in direct sequence CDMA. We project the received signal to the time-frequency domain using an adaptive Gabor transform. Then we detect the jammer using the SG-CA CFAR detector and proceed to excise it. This algorithm provides about 10% improved SER (symbol error rate) ratio compared to non-optimized Gabor transform. |
