建立基于电子鼻的猪肉冷冻储藏期和冻藏前冷藏时间的无损检测方法。5 cm见方冻藏猪肉置于100 m L烧杯中以保鲜膜密封,采用PEN3型电子鼻对烧杯中上层气体进行检测。检测数据分别采用主成分分析和重复测量的方差分析了解多元数据的分布特...建立基于电子鼻的猪肉冷冻储藏期和冻藏前冷藏时间的无损检测方法。5 cm见方冻藏猪肉置于100 m L烧杯中以保鲜膜密封,采用PEN3型电子鼻对烧杯中上层气体进行检测。检测数据分别采用主成分分析和重复测量的方差分析了解多元数据的分布特点。进而利用线性判别和基于多层感知器的神经网络算法建立猪肉冷冻储藏期和冻藏前冷藏时间的预测模型。结果表明,冷冻储藏不同时间及冻藏前冷藏不同时间的猪肉样品的电子鼻检测数据存在显著差异;通过神经网络算法建立的预测模型对猪肉冷冻储藏期和冻藏前不同冷藏时间具有更好预测能力。该检测方法操作简便,快速高效可实现无损现场检测,为加强冷冻肉品监督管理提供了一种简便可行的检测技术。展开更多
This paper introduces a new concept of ＂State Representation Methodology （SRM）＂ which is a kind of bridge condition assessment method for structural health monitoring system （SHM）. There are many methods for sys...This paper introduces a new concept of ＂State Representation Methodology （SRM）＂ which is a kind of bridge condition assessment method for structural health monitoring system （SHM）. There are many methods for system identification from the simplicity comparison of damage index to the complicated statistical pattern recognition algorithms in SHM. In these methods, modal analysis and parameters identification or many defined indices are common-used for extracting the dynamic or static characteristics of a system. However, there is a common problem： due to the complexity of a large size system with high-order nonlinear characteristics and severe environment interference, it is impossible to extract and quantify exactly these modal parameters or system parameters or indices as the feature vectors of a system in damage detection in an easy way. The SRM considered a more general theory for the non-parametric description of system state.展开更多
文摘建立基于电子鼻的猪肉冷冻储藏期和冻藏前冷藏时间的无损检测方法。5 cm见方冻藏猪肉置于100 m L烧杯中以保鲜膜密封,采用PEN3型电子鼻对烧杯中上层气体进行检测。检测数据分别采用主成分分析和重复测量的方差分析了解多元数据的分布特点。进而利用线性判别和基于多层感知器的神经网络算法建立猪肉冷冻储藏期和冻藏前冷藏时间的预测模型。结果表明,冷冻储藏不同时间及冻藏前冷藏不同时间的猪肉样品的电子鼻检测数据存在显著差异;通过神经网络算法建立的预测模型对猪肉冷冻储藏期和冻藏前不同冷藏时间具有更好预测能力。该检测方法操作简便,快速高效可实现无损现场检测,为加强冷冻肉品监督管理提供了一种简便可行的检测技术。
文摘This paper introduces a new concept of ＂State Representation Methodology （SRM）＂ which is a kind of bridge condition assessment method for structural health monitoring system （SHM）. There are many methods for system identification from the simplicity comparison of damage index to the complicated statistical pattern recognition algorithms in SHM. In these methods, modal analysis and parameters identification or many defined indices are common-used for extracting the dynamic or static characteristics of a system. However, there is a common problem： due to the complexity of a large size system with high-order nonlinear characteristics and severe environment interference, it is impossible to extract and quantify exactly these modal parameters or system parameters or indices as the feature vectors of a system in damage detection in an easy way. The SRM considered a more general theory for the non-parametric description of system state.
