一种加窗数据降维方法在食品品质分析中的应用

加窗处理方法是在应用近红外光谱分析技术对食品品质进行分析、建立数学预测模型的过程中，选用适当窗函数对原始光谱数据施行降维、降噪等操作，从而节约计算机内存、减少运算量，并提高模型的预报精度。对大米近红外光谱数据，分别采用不加窗处理和加窗处理措施建立了大米蛋白质含量的预测模型。经实验，该方法能有效地压缩原始数据、降低噪声影响，明显地缩短了计算机的运行时间，使模型评价指标如平均相对误差、平均偏差和预测标准差有所降低，使预测值和化学参比值的相关系数(R^2)由0．9204提高到0．9526。

Application of Cutting Down Method of Data with Adding Windows in Food Quality Analysis

Adding windows method means that,in the process of using near infrared spectroscopy to analyze food quality and establish mathematic model,it is necessary to cut down the primary data and noise etc.by choosing suitable windows function in order to saving computer memory,reducing calculation and improving forecasting precision of the model.For rice near infrared spectrum data,the forecasting models of rice protein content were established with respectively no-adding windows and adding windows method.The experiment results showed that the adding windows method can not only compress the primary data,lower the influence of the noise effectively,shorten the calculating time of computer obviously,but make the evaluating indexes such as average relative error,average deviation and forecasting standard error lower,and improve the relative coefficient(R~2) between forecasting value and chemical comparative value from 0.920 4 to 0.952 6.