Minor Component Analysis Based Design of Low Pass and BandPass FIR Digital Filter Using Particle Swarm Optimization and Fractional Derivative

In this work, a new approach lean on minor component analysis (MCA) neural learning and fractional derivative (FD) is introduced for the design of digital finite impulse response (FIR) filters. In this method, design problem is modeled as summation of integral error in passband and stopband region in term of polyphase components (PCs) of a FIR filter in frequency domain, which is solved by an efficient machine learning algorithm called minor component analysis (MCA) neural learning. For more accurate frequency response, fractional derivative is applied at a reference point in passband, and the resulted fractional derivative constraints (FDCs) are optimized by particle swarm based techniques, using an objective function formulated as summation of maximum error in pass band and stop band region and stopband attenuation (in magnitude) of a FIR filter. The comparative study with recently published results evidence the impact of proposed method.