A sensor array composed of ion electrodes including 2 glass electrodes, 3 liquid-membrane electrodes and 7 insoluble salt electrodes was built. Before detection, the working electrodes were activated as required in activate fluids, and the stability states of sensors were analyzed in deionized water. Beef samples were evaluated after all working electrodes stabilized. The response signals from the samples were recorded by an electrochemical workstation and used as the evaluation results. A beef taste sensory evaluation criterion was built and used into sensory evaluation of beef samples. The samples were scored with quality grades according to this criterion, and the results were compared with the results of the sensor array in evaluation of beef broth samples. The evaluation results were processed by principal component analysis and used to build a beef taste quality evaluation model based on artificial neural networks. Tests show this model has an accuracy of 90% in classification of beef taste quality grades.
Keywords: Beef; Evaluation; Neural networks; Sensor array; Taste.
Copyright © 2017 Elsevier Ltd. All rights reserved.