中文摘要：

为科学客观地评估鲜叶收购价格,应用近红外光谱技术结合人工神经网络方法和联合区间偏最小二乘法,建立了三种鲜叶收购价格预测模型并比较了预测效果.应用联合区间偏最小二乘法筛选最佳光谱区间为5 750-6 000cm^-1,7 750-8 000cm^-1,8 250-8 500cm^-1,8 500-8 750cm^-1,9 500-9 750cm^-1和9 750-10 000cm^-1,并对上述光谱进行主成分分析.前5个主成分累计贡献率为99.87%,并以此为输入值建立收购价格人工神经网络预测模型（R^2=0.968 7,RMSEP=4.625）.模型预测结果优于全波长人工神经网络模型（R^2=0.855 1,RMSEP=5.218）和联合区间偏最小二乘法模型（R2=0.581 6,RMSEP=25.433）的预测结果.近红外光谱技术结合人工神经网络和联合区间偏最小二乘法,能够快速、准确、客观的评估鲜叶收购价格,有利于统一鲜叶收购价格标准,有效地减少纠纷。

英文摘要：

Near infrared spectroscopy combined with the back propagation artificial neural network algorithm and the synergy interval partial least square algorithm was used to evaluate the purchasing price of fresh tea leaves. The nearinfrared spectra regions of 5 750 cm^-1 to 6 000 cm^-1, 7 750 cm^-1 to 8 000 cm^-1 , 8 250 cm^-1 to 8 500 cm^-1 , 8 500 cm^-1 to 8 750 cm^-1 , 9 500 cm^-1 to 9 750 cm^-1 and 9 750 cm^-1 to 10 000 cm^-1 were selected to establish a model by using the synergy interval partial least square algorithm. The first five principal components that explained 99.87 - of the variability of the selected spectral data were used to build tea leaves＇ purchasing price model with the back propagation artificial neural algorithm. The performance of this model （R^2 , 0. 968 7; RMSEP, 4. 625） was superior to those of the back propagation artificial neural model （Re , 0.8551 ;RMSEP, 5. 218） and the synergy interval partial least square model （R^2, 0. 581 6; RMSEP, 25. 433）, The near infrared spectroscopy combined with the synergy interval partial least square algorithm and the back propagation artificial neural network algorithm could be used to evaluate the price of Enshi Yulu tea leaves accurately, quickly and objectively.