基于近红外光谱的SSA-RELM的菠萝含水率快速检测

目的：建立快速无损检测菠萝含水率的方法。方法：提出一种基于连续投影法的特征波长选择和麻雀搜索算法（SSA）优化正则化极限学习机（RELM）的菠萝含水率检测模型。针对菠萝近红外光谱数据具有维度高、冗余信息多的特点，分别对比连续投影法、主成分分析法和全波段等筛选特征波长的结果，确定菠萝近红外光谱特征波长筛选方法；针对RELM模型性能受其输入层权值和隐含层偏置的影响，运用麻雀搜索算法优化RELM模型的输入层权值和隐含层偏置，提出一种基于麻雀搜索算法改进正则化极限学习机的菠萝含水率检测模型。结果：与遗传算法改进正则化极限学习机（GA-RELM）、粒子群算法改进正则化极限学习机（PSO-RELM）和RELM相比，基于麻雀算法改进正则化极限学习机（SSA-RELM）的菠萝含水率检测模型的检测精度最高。结论：麻雀搜索算法优化RELM模型可以有效提高RELM模型的菠萝含水率检测精度。

Rapid detection of moisture content of pineapple based on near infrared spectroscopy and SSA-RELM

Objective: A method for a fast and non-destructive detection of pineapple moisture content was established. Methods: A novel detection model of pineapple moisture content was proposed based on continuous projection feature wavelength selection and Sparrow search algorithm. Firstly, according to the characteristic of pineapple NIR data with high dimension and redundant information, the results of feature wavelength selection such as successive projections algorithm, principal component analysis and full-band were compared, the selection method of characteristic wavelength of pineapple near infrared spectrum was determined. Secondly, considering that the performance of RELM model was affected by the selection of input layer weight and hidden layer bias, the sparrow search algorithm was used to optimize the input layer weight and hidden layer bias of RELM model, a novel pineapple moisture content detection model based on RELM model improved by sparrow search algorithm was proposed. Results: compared with GA-RELM, PSO-RELM and RELM, the detection model based on SSA-RELM had the highest detection accuracy. Conclusion: RELM model is optimized by sparrow search algorithm can effectively improve the detection accuracy of RELM model .