Application of least squares support vector machine in soft sensor of traditional Chinese Medicine extraction

Abstract

Aiming at the difficult measurement problem of the extraction rate for plants and herbs with the ultrasonic wave technology, the influence of the various factors on the extraction rate in the ultrasonic extraction process is analyzed and the dynamic process variables which is easily measured and can affect the extraction rate is ensured in this paper. A soft sensor model between the easily measured variables and the ones to be measured is established with the Least Squares Support Vector Machine (LS-SVM) method. Using the optimized model, the impact of process parameters on the extraction rate in the extraction process of Chinese medicine is predicted and analyzed. The learning performance and generalization capability of the model are verified. The conclusion that the extraction temperature has an impact on the extraction rate of the traditional Chinese medicine can be drawn. Finally, the experimental results show that the LS-SVM method is suitable for data modeling of small sample data and characterizes by the quicker calculation speed and stronger generalization ability. The soft sensor model which is established with the LS-SVM method has achieved more accurate prediction on extraction rate of the traditional Chinese medicine.