The soluble solids content (SSC) is a key parameter to affect the quality of Korla fragrant pears. Having reliable and rapid measurement of the SSC is crucial for producers and technicians. However, the equipment necessary to meet these requirements is frequently complex and expensive. This paper developed a portable nondestructive SSC detector, primarily composed of a handheld near-infrared (NIR) spectrometer in the wavelength of 900–1700 nm, a display, a lithium battery, and a Raspberry Pi board. In addition, the whole detector was only about 425 g in total. To test the performance of the prototype, we chose 100 Korla fragrant pears as our fruit samples. Different spectral pre-processing methods were combined with principal component regression (PCR) and partial least squares regression (PLSR) to obtain accurate models. To further simplify the model, the synergy interval (Si), genetic algorithm (GA), and random frog (RF) were used to choose characteristic wavelengths. Laboratory tests show that compared with PLSR and PCR based on the full-spectrum, the prediction models obtained after conducting RF selection have yielded satisfactory results. The number of wavelength features was reduced from 228 to 10, and the R² of the model was improved from 0.942 to 0.966. The mean relative error rate in the field test was 1.41%, indicating that the developed SSC detector was also effective in the field. It can be verified that the NIR diffuse reflectance spectroscopy is a reliable tool for the nondestructive measurement of SSC in Korla fragrant pears. The appropriate pretreatment methods and selection of characteristic wavelengths can increase NIR spectroscopy accuracy in actual measurement.

可溶性固形物含量（SSC）是影响库尔勒香梨品质的关键参数。对SSC进行可靠，快速的测量对于生产者和技术人员而言至关重要。然而，满足这些要求所需的设备通常是复杂且昂贵的。本文开发了一种便携式无损SSC检测器，主要由波长为900-1700 nm的手持式近红外（NIR）光谱仪，显示器，锂电池和Raspberry Pi板组成。另外，整个检测器总共仅约425g。为了测试原型的性能，我们选择了100个库尔勒香梨作为我们的水果样品。将不同的光谱预处理方法与主成分回归（PCR）和偏最小二乘回归（PLSR）相结合，以获得准确的模型。为了进一步简化模型，使用了协同区间（Si），遗传算法（GA）和随机青蛙（RF）来选择特征波长。实验室测试表明，与基于全谱的PLSR和PCR相比，进行RF选择后获得的预测模型取得了令人满意的结果。波长特征的数量从228个减少到10个，R该模型的^2个从0.942提高到0.966。现场测试的平均相对误差率为1.41％，表明开发的SSC检测器在现场也有效。可以证明，近红外漫反射光谱法是无损测量库尔勒香梨中SSC的可靠工具。适当的预处理方法和特征波长的选择可以提高实际测量中的近红外光谱精度。