【摘要】 针对近红外光谱在线监测软件算法模型扩展功能受限和对外服务能力受限的问题,论文设计出一种预测精准且开放服务能力强的近红外在线葡萄糖浓度监测系统。首先,对原始的光谱数据进行预处理,对比多种奇异样本剔除算法以及区间选择算法,从而优选出最佳区间并建立葡萄糖浓度的定量分析模型。然后,将优选模型嵌入到基于java和websocket的后台检测接口中,解决模型嵌入问题,采用微信小程序作为前端展示平台可以很好地解决平台不适应问题。经过测试分析,所嵌入的优选模型具有更好的预测精度,所设计系统可以跨平台多终端远程访问,实时监测效果良好。更多还原

【Abstract】 Aiming at the problem that the extended algorithm function of the near-infrared spectroscopy online monitoring software algorithm model is limited and the external service capacity is limited,the paper designs a near-infrared online glucose concentration monitoring system with accurate prediction and open service capability. Firstly,the original spectral data is preprocessed,and various singular sample culling algorithms and interval selection algorithms are compared to optimize the optimal interval and establish a quantitative analysis model of glucose concentration. Then,the preferred model is embedded into the background detection interface based on java and websocket to solve the model embedding problem. The WeChat applet can be used as the front-end display platform to solve the platform incompatibility problem. After testing and analysis,the embedded preferred model has better prediction accuracy,and the designed system can be remotely accessed across platforms and terminals,and the real-time monitoring effect is good.更多还原