The exact regulation of dry matter, energy and ingredients in fodder rations provides a large advantage in order to optimize an economical animal nutrition. Feed mixer wagons are used to feed Gras and Maize silage together with other components. It can be used in combination with a transponder system for feed concentrate as well as for feeding of a total mixed ration. The online measurement system based on NIR-spectrometric sensors to measure DM-content and other nutrients should result in a more precise adjustment of the ration. Main object of the project “SenToMiRa” is a nutrient optimized total mixed ration for feeding dairy cattle by installation of a near infrared sensor (MUT-Group) on the material flow at a self-propelled feed mixer wagon (STRAUTMANN).

To calibrate the measurement system spectra of many different silage samples are taken offline and analysed by wet chemistry. The goal is a calibration model which can be used in the online measurement system. The NIRS sensor and the spectrometer have been integrated to the loading system of the feed mixer wagon. The first calibration model is the start point for the online measurement at the Feed mixer wagon. The model has to be improved by testing more samples of different silos. While loading the feed crop a part of the mass flow has to pass directly in front of the NIRS detector head. To ensure a constant reflectance measurement it is important to have self cleaning glass and no fodder package in front of the detector head. Damages cause by hard particles like stones are prevented by a resistant sapphire glass.

For a precise measurement of fodder it is important to have calibration with high accuracy. To get a good calibration model 400 - 500 samples are needed. The samples are measured with the near infrared sensor and calibrated by the wet chemistry analyzes. For the calibration of the NIRS the samples were measured with an offline method by using a developed turntable.

The samples are given into a plate which is turned directly on top of the detector head.

This gives the opportunity to calculate an average spectra of the sample. Reference measurements of dry weight, nutrients and energy content with wet chemistry analysis in the laboratory complete the components for the calibration. The wet chemistry analyses are performed by a certificated laboratory. By the measurements of each silo surface different samples were taken.

Figure 2 shows for example the comparison of the values measured by near infrared spectroscopy estimator function with the values from the wet-chemical analysis of reference. The solid line represents the regression. Figure 7 show the model calibration of DM content in grass silages. The first calibration of grass silage with dry matter shows very good results. The root mean squared error of prediction (RMSEP) is located at 3,194. The range/standard error of performance (SEP) has a value of 10,738. The displayed calibration model consists of 98 samples. The samples are evenly distributed over a range of DM 22 – 53%.

In the presentation the actual models for DM and nutrients will be shown and explained as well as the installed NIRS system at the test feed mixer wagon.