Designed strength identification of concrete by ultrasonic signal processing based on artificial intelligence techniques

This paper presents a pattern recognition method to identify the designed strength of concrete by evidence accumulation based on artificial intelligence techniques with multiple feature parameters. Concrete specimens in this experiment, which were designed to have the strengths of 180, 210, 240, 300, and 400 kg/cm/sup 2/, respectively, have been considered. Variance, zero-crossing, mean frequency, autoregressive (AR) model coefficients, and linear cepstrum coefficients are extracted as feature parameters from ultrasonic signals of concretes. Pattern recognition is carried out through the evidence accumulation procedure using the distances measured with reference parameters. A fuzzy mapping function is introduced to transform the distance for the application of the evidence accumulation method. Results are presented to support the feasibility of the suggested approach for concrete pattern identification.