基于K-近邻算法的物料含水率测量标准装置设计

目的 针对烘箱法耗时长且不适用于生产现场的问题，围绕短时干燥法设计物料含水率测量标准装置。方法 装置主要包括移动底盘、箱体、水分分析仪、可充电电源、配电板、工业触屏电脑以及上位机软件。建立基于K-近邻（K-Nearest Neighbor）算法的物料含水率预测模型，并通过实验获得预测结果与烘箱法测量结果之间的修正关系。以抽取自卷烟厂的烟草样品为例，对装置进行测量实验。结果 当设置测量时间为30min时，含水率测量结果的扩展不确定度低于0.5%。结论 该装置可代替烘箱法对物料含水率进行监控，也可以用于在线水分仪的原位校准。

Standard Device Design for Matter Moisture Content Measurement Based on KNN Algorithm

The work aims to design a standard device for matter moisture content measurement based on the short time drying method, in order to solve the problems of poor timeliness and improper installation of standard oven on production site. The device was mainly composed of a mobile chassis, a box, a moisture analyzer, a rechargeable battery, a power distribution board, an industrial touch screen computer and PC software. The prediction model of matter moisture content based on K-nearest neighbor (KNN) algorithm was established, and the correction relationship between the prediction results and the measurement results of standard oven method was obtained. The measurement experiment was carried out to the device with tobacco samples provided by the cigarette factory. When the measurement time was 30 min, the expanded uncertainty of the moisture content measurement result by the device was no more than 0.5%. This device can replace the oven method to monitor the moisture content of matters, and can also be used for the in-situ calibration of online moisture analyzer.