基于PSO-SVR算法的牛乳电导率近红外回归模型研究

摘要

The rapid and accurate detection of milk electrical conductivity is of great significance to the healthy development of the dairy farm.A novel rapid detection method using near infrared spectroscopy combined with chemometrics was proposed in this paper. For the NIR spectral data of 90 farm milk samples, the support vector regression models were built by using genetic algorithm and particle swarm algorithm respectively, the results show that the PSO-SVR model has better performance and higher prediction preci-sion compared with the GA-SVR model and the traditional PLS model.The near infrared model based on PSO-SVR algorithm can be applied to the rapid and accurate measurement of the milk electrical conductivity.%牛乳电导率值是判断奶牛是否感染乳腺炎的重要依据，其快速、准确测量对奶牛养殖业的健康发展具有重要意义。本文提出一种近红外光谱结合化学计量学快速测量牛乳电导率的新方法。针对90个牛乳样本的近红外光谱，建立电导率值的支持向量机回归模型，分别采用遗传算法和粒子群算法进行模型参数寻优，结果表明基于粒子群寻优算法所建立的牛乳电导率值预测模型，相比GA-SVR模型和传统PLS模型具有更好的性能指标，预测精度更高，可以应用于牛乳电导率的快速、准确测量。