NPK Fertilizer Calibration Demonstration

氮磷钾肥料实验报告

Basic Operating Principle of NIRS近红外的基础原理

In fact, different fertilizers have different spectrums. Different composition contents also have different spectrums.事实上，不同的肥料有不同的光谱。不同的成分含量也有不同的光谱。

Therefore, we can regard different spectrums as independent variable (Like X) and regard different composition contents as dependent variable (Like Y). Based on mathematical analysis (like PLS, MLR, PCR and so on ), we can set up a specific calibration model:

因此，我们可以将不同的光谱作为自变量(如X)，将不同的成分含量作为因变量(如Y)。通过数学分析(如PLS、MLR、PCR等)，我们可以建立一个具体的校准模型:

Y = a1*X1 + a2*X2 +…+an*Xn + b.

Corona Family Introduction       Corna系列仪器介绍

ZEISS Corona System is a robust device without moving parts. It can be mounted on mobile machinery for online measurement of solid sample quality, such as a grain harvester or over a moving cement powder belt.蔡司Corona系统是一种没有运动部件的耐极端条件的光谱仪装置。它可以安装在移动机械上在线测量固体

样品的质量，如谷物收割机、移动传送带或其他生产管道和罐体上。

              平面光栅光谱仪的元件

Samples - Calibration Results

样品 - 校准模型

These are 10 samples from April 16th 2pm to April 17th 8am. And they are from only one production line GP1. We collected 5 spectrums for each sample. This model is based on 50 spectrum-chemical values.

这是4月16日下午2点到4月17日上午8点的10个样品。它们只来自一条生产线GP1。我们为每个样品收集了5个光谱。该模型基于50个光谱化学值。

Samples - Calibration Results

样本-校正结果

These are 14 samples from April 16th 2pm to April 17th 8am from two production lines GP2 and GP3. We collected 3 spectrums for each samples. This model is based on 42 spectrum-chemical values.   

这是4月16日下午2点到4月17日上午8点的14个样品，来自GP2和GP3两条生产线。我们为每个样品收集了3个光谱。该模型基于42个光谱化学值。

The first Samples – Used for Model Cross Validation

第一种样品-用于模型交叉验证

Chemical laboratory data: nitrogen is 13.77, phosphor is 14.41, potassium is 13.44, the moisture is 1.24.   

化学实验室数据:氮为13.77，磷为14.41，钾为13.44，水分为1.24

NIR predicted value

The second Samples - Cross Validation

第二种肥料样品-交叉验证

Chemical laboratory data: nitrogen is 16.37, phosphor is 20.06, moisture is 1.01.

化学实验室数据:氮为16.37，磷为20.06，水分为1.01。

NIR predicted value

Statistical Modelling Introduction (GD & ND) – Data and Model Quality Indicators

统计模型介绍(GD & ND) -数据和模型质量指标

GD: Global Distance

ND: Neighborhood distance

How to validate the current model?

如何验证当前模型?

How to validate the current model? Data correlation

如何验证当前模型?数据相关性

R2=0.81 SEP=0.11

Corona process – inline – for production Increase production efficiency提高生产效率

The Corona process is installed in a distance above the material and can be fitted with an “airbrush” to prevent dust reaching the measurement window.

CoronaProcess安装在材料上方一定距离的地方，可以安装一个“airbrush”，以防止灰尘影响测量窗口。

-No impact of corrosive dust.

-No abrasion by granules

-Shock/vibration resistant

-Ingress protection

-腐蚀性粉尘对其无影响。

-无需颗粒磨损

-耐冲击/振动

-防护等级

Quality Assurance Monitoring in Production   生产过程中的质量保证监控

Process Quality Control   过程质量控制，可实现现场报警提示。

Example*: Orange: not enough sample in the dish=> Warning

GREEN: sufficient sample in the dish=> NO Error

RED: no sample in the dish=> Error, Optical and Acoustical

*Realization via Beckhoff terminal and light tower

Conclusion   总结

We have shown that, within a short time and comparatively modest effort, a useful analytical method can be developed in the laboratory using Corona extreme with Turnstep.

An offline installation in the laboratory or at-line in the plant allows fast determination of multiple parameters simultaneously.

在实验室离线检测或者生产在线检测，可同时快速测定多个理化指标。

All installation options (inline, at-line,

offline) allow monitoring from the control

room and the laboratory (TV splitter) of:

-    Moisture %

-    N-content

-    P-content

-    K-content

所有安装都可实现安装HMI人界交互界面，实时监控质量数据：水分、氮、磷、钾等。                                                                                                             

山东智惠仪器有限公司    

更多实验详情，我们愿与您分享，欢迎联系我们！