Partial least square regression techniques in obtaining measurements of one or more polymer properties with an on-line nmr system

A method for predicting a value of a property of a polymer sample in real time from an on-line production process, said method comprising the steps of:

• storing a plurality of models wherein at least two of the models are useful for predicting a value of a property of interest of a production polymer sample and at least one of the models is useful in identifying which of the at least two models more accurately predicts the value of the property of interest;
• providing a production sample having an unknown value of said property of interest;
• applying a base magnetic field to the production sample to effect precession of nuclei of the production sample;
• modifying the precession;
• receiving a resulting relaxation signal representative of a free induction decay of the nuclei of the production sample;
• digitizing said relaxation signal;
• applying an iterative technique to derive respective component curve equations from said relaxation signal;
• calculating respective component curve equation coefficients;
• utilizing the coefficients, or information based on the coefficients, and the at least one model in order to select one of the at least two models;
• utilizing the coefficients, or information based on the coefficients, and the selected one of the at least two models in order to predict the value of the property of interest, wherein at least one of said two models is generated by a partial least square regression technique.