LAUSR

The aims of this work were to develop multi-element viscoelastic models for beef and apply them to detect total volatile basic nitrogen (TVB-N) content for freshness evaluation. The deformation data were collected by a viscoelasticity detection system that employed the airflow and laser technique. Then, TVB-N contents were measured to determine the freshness of samples during storage. A universal global optimization (UGO) algorithm was applied to fit the deformation data. Various multi-element viscoelastic models including the Burgers, six-element and eight-element models were built using the obtained fitting parameters, and different viscoelastic parameters representing the degree of beef spoilage were obtained. All the viscoelastic parameters of each multi-element model and parameter combinations of the selected six-element model were employed to build mathematical models for predicting TVB-N content by support vector machine regression (SVR). In comparison, the six-element model with all the viscoelastic parameters performed the best and was determined to predict TVB-N content with correlation coefficient in the prediction set (RP ) of 0.891 and root mean squared error in the prediction set (RMSEP) of 1.467 mg/100 g. Based on the results of parameter combinations, combination (E2 , E3 , E1 , η1 , η2 ) from the six-element model performed the best, which was comparatively inferior to all the viscoelastic parameters of the six-element model. Results demonstrated that it was possible to predict TVB-N content for freshness evaluation by applying method of developing multi-element model based on the viscoelasticity with chemometrics.