Hardware accelerated Wavelet Transform and de-noising for pattern recognition

Wavelet Transform is a widely used tool in signal processing which helps in localizing a signal in both time and frequency domain. This is in contrast to FFT which can only resolve a signal in frequency domain. Additionally, Wavelet Transform has been widely employed by electrical machines researchers for fault diagnosis and prognosis especially as means to extract features for classification purposes. The problem though with both machine learning algorithms and Wavelet Transform is that they require extensive computational resources which are not available in environs where electrical machines are most often placed. Additionally, these environments are plagued by noise which considerably affects the quality of input signal i.e., current. Other issues related to on-board system is the associated cost of diagnosis hardware. Concretely, the algorithm for Wavelet Analysis of a signal should be capable of producing reasonable results using cost effective hardware which comes with the compromise of lesser resolution, limited memory and low power consumption. In this proposed work, implementation scheme of Wavelet Transform on low cost PSoC3 hardware is presented which tightly integrates the FPGA blocks and DSP processor on the chip with microcontroller core to yield a fast and robust hardware accelerated mechanism to compute Wavelet Transform, Wavelet thresholding for denoising and classification.