摘要
通过自行研制的一套禽蛋裂纹检测装置,采集并分析敲击鸡蛋产生的响应信号,检测裂纹鸡蛋。采用基于归一化最小均方算法的自适应滤波器对信号进行去噪处理。结果表明:经自适应滤波后,敲击响应信号的分辨率和灵敏度均有显著提高。提取经滤波去噪后的鸡蛋敲击响应信号功率谱的5个特征参数,作为误差反传人工神经网络模型的输入向量进行判别。判别模型对实验鸡蛋的交互验证训练集和独立样本预测集的判别率均为97%。
A system based on acoustic resonance was developed for eggshell crack detection. It was achieved by analysis of the acoustic frequency response of eggshell excited with a light mechanical. Least mean square (LMS) adaptive filtering were attempted to remove the noise from the response signal. Some characteristics were extracted from excitation resonant frequency and served as input vectors of discrimination model. Back propagation artificial neural network (BP-ANN) model was employed to discriminate intact eggs and cracked eggs. This method allows a crack detection level of 97% in calibration and prediction set respectively. This work shows that acoustic resonance system combined with adaptive filters has a significant potential in cracked eggs detection.
出处
《食品科学》
EI
CAS
CSCD
北大核心
2010年第2期199-202,共4页
Food Science
基金
"十一五"国家科技支撑计划项目(2006BAD11A12)
关键词
鸡蛋
裂纹
声音响应信号
归-化最小均方算法
误差反传人工神经网络
egg
crack
acoustic resonance
normalized least mean square (NLMS)
back propagation artificial neuralnetwork (BP-ANN)