NIR cheese analysis using ANN calibrations

Analysis of moisture, fat, protein salt and pH is of great importance to cheese manufacturers and can be measured accurately by NIR. By using ANN technique it was possible to develop calibrations for these parameters across a wide variety of cheese types including cottage cheese, mozzarella and grated hard cheeses.

By accurately controlling these constituents, the producer can experience significant savings and ensure product quality. Availability of fast results allows for process optimizations and to avoid costly mistakes and potential penalties.

The DA 7250 is a proven NIR instrument designed for use in the food industry. Using novel diode array technology it performs a multi-component analysis in only 6 seconds. It has been used in cheese manufacturing for years, and to improve its usefulness for cheese manufacturers, a project was set up to develop updated calibrations.

Around 10,000 cheese samples from processing plants in North America and Europe were collected and combined to serve as NIR calibration set. The calibration set included a large variability of in-process and finished cheeses, from mozzarella and soft cheeses to hard cheeses. Collected during various process steps such as pressing stage, brining, ripening and finished cheese.

Several calibration techniques were evaluated and in the end calibrations models based on Artificial Neural Networks (ANN) were selected. With this technique it was possible to combine all cheese types into one calibration, with excellent accuracies for all parameterers.

An application note which describes the cheese application in more detail is available for download here.

Learn more about the DA 7250 here.

Read about Perten's other instruments for dairy products here.