腐败肉光谱在线检测及剔除系统研究

Online detection and elimination system for spoilage meat using spectrum

目的提供一种基于可见/近红外光谱技术的腐败猪肉在线检测及剔除系统,实现自动化检测。方法构建由同步模块、检测模块和剔除模块组成的检测系统。肉样随传送带运动,同步模块控制检测探头的调整,保证探头到肉样表面的距离,并使其与肉的传送同步进行;检测模块用于提取光谱并进行分析,得出是否腐败的结果;之后由剔除模块对腐败样本进行剔除。将光谱数据与理化值联立采用多元散射校正与主成分分析法进行光谱预处理,建立了基于Fisher算法的判别模型。结果系统信噪比>100:1,在线判别准确率达到87.4%。结论该系统可以保证光谱检测的稳定性,能实现腐败肉的在线检测和剔除动作,为实际生产应用创造了条件。

Objective To detect and eliminate the spoiled pork on-line automatically by providing a system based on visible/near infrared spectroscopy. Method The online examination system was mainly composed of synchronization module, test module and elimination module. Specimens were moved with conveyor belt, then synchronous module adjusted testing probe to ensure that the probe moves with specimen together; detection module was used to extract and analyze the spectrum and to gain the result of the discriminant; in the end, elimination module eliminated spoiled samples. To combine spectral data, the physical, chemical values and pretreated spectra with the multiple scattering correction and principal component analysis （PCA）, discriminant model based on Fisher algorithm was established. Results Signal-to-noise ratio of system was more than 100：1, and the online discriminant accuracy reached 87.4%. Conclusion The experimental results show that the system can guarantee the stability of the spectrum detection, realize on-line detection and eliminate spoiled meat, which creating the conditions for the actual production.