论文部分内容阅读
由超声脉冲法获得4 种检测信息——声速、波幅、频率和波形,从中提取3 个特征量:平均声速、声速的离差系数和波幅的离差系数,利用模糊分析方法和人工神经网络相结合对 G R C、加气混凝土、空心砖、实心砖、砌块 5 种墙体材料进行种类识别,形成一个智能化的、具有类似人脑功能的系统.经现场实测证明,判别的准确率达95% 以上。
Four kinds of detection information—sound speed, amplitude, frequency, and waveform—were obtained by the ultrasonic pulse method. Three feature quantities were extracted: the average speed of sound, the coefficient of variation of the speed of sound, and the coefficient of dispersion of the amplitude, using fuzzy analysis methods and artificial neural networks. Combined with the recognition of five types of wall materials such as GRC, air-entrained concrete, hollow brick, solid brick, and block, an intelligent system with human brain function is formed. The field test proved that the accuracy of the discrimination reached more than 95%.