| 摘要: |
| 【目的】结合遗传算法和最小二乘支持向量机(GA-LSSVM)，优化苹果糖度近红外光谱检测的数学模型，提高模型的检测精度和稳定性。【方法】在GA-LSSVM模型建立过程中，采用遗传算法自动获取最小二乘支持向量机的最优参数。【结果】相比于偏最小二乘法(PLS)、传统最小二乘支持向量机(LSSVM) 和遗传偏最小二乘法(GA-PLS) 数学模型，GA-LSSVM法建立的模型预测效果最优，模型的相关系数为0.94，预测均方根误差为0.32 °Brix。【结论】GA和LSSVM相结合的优化方法在提高苹果糖度近红外光谱检测精度和稳定性方面是可行的。 |
| 关键词: 苹果 糖度检测 近红外光谱 遗传算法 最小二乘支持向量机 |
| DOI： |
| 分类号: |
| 基金项目:科技部农业科技成果转化项目(2011GB2C500008)；赣鄱英才555工程领军人才培养计划项目 |
|
| GA-LSSVM based near infrared spectroscopy detection of apple sugar content |
|
|
| Abstract: |
| 【Objective】The objective of the present research was to optimize the detection of sugar content in apples for improving the detection precision and robustness using near infrared spectroscopy,combined with genetic algorithms and least squares support vector machine(GA-LSSVM).【Method】In the process of establishing GA-LSSVM model,GA method was used to select the optimal parameters of LSSVM automatically.【Result】 Compared with partial least squares(PLS) model,GA-PLS model and LSSVM model,GA-LSSVM model was more accurate than others.The correlation of predictive model(Rp) was 0.94,and the root mean square error of prediction(RMSEP) was 0.32 °Brix.【Conclusion】It was feasible to improve the precision of near infrared spectroscopy detection of apple sugar content by the combination GA and LSSVM. |
| Key words: apple sugar content detection near infrared spectroscopy genetic algorithms least squares support vector machines |
