Abstract: Fuzzy entropy is used to determine the uncertainty of characteristic parameters of rub-impact acoustic emission (AE) signal in the different rub-impact modes. Average signal level, amplitude, dynamic amplitude range and energies of four nodes reconstruction signals decomposition by wavelet packets are chosen as the characteristic parameters of recognition of AE signal. The membership function of each characteristic parameter belonged to different rub-impact mode based on Gaussian model is obtained from the training samples respectively. Subsequently the fuzzy relation matrix of characteristic parameters and modes is calculated with membership functions. According to different effectiveness in recognizing AE signal by the characteristic parameters, an algorithm based on fuzzy entropy is presented to calculate effectiveness coefficient. The integrated evaluate fuzzy sets is calculated with the new fuzzy relation matrix modified by effectiveness coefficient, and the mode which has the most degree of membership is chosen as the recognition results. The experiments indicate that the integrated fuzzy evaluation is effective in analysis and recognition of AE, and the differences of parameter effectiveness can be used to improve recognition efficiency.

Key words: Acoustic emission, Effectiveness coefficient, Fuzzy entropy, Signal recognition