A NOVEL EMD-IABC BASED DE-NOISING FOR GRAIN IMPACT SIGNAL

To remove noise from the grain impact signal, a de-noising method that incorporates empirical mode decomposition (EMD) and artificial bee colony (ABC) is proposed. EMD decomposes a signal into intrinsic mode functions (IMFs) that are with different frequencies. The de-noising signal is obtained by superposition of the IMFs processed with a threshold set. The ABC optimises a threshold set for its optimal value. A search strategy and a probability model are introduced to the ABC algorithm, namely, the improved ABC (IABC), enhancing its exploitation and exploration capacity. Simulation results show that under 1-dB Gaussian white noise, the proposed method obtains maximum SNR that rises by 5.68% and 6.93%, and minimum RMSE that falls by 5.11% and 4.37%, compared with EMD-ABC and EMD-PSO. As the noise level rises, the proposed method still maintains a good de-noising effect. Practical application results show that the proposed method is effective in removing noise from grain flow signals.