In this paper, a new method for high precision and wide range measurement of chemical oxygen demand (COD) based on ultraviolet absorption spectroscopy without reagent is proposed. The reasons for limiting measurement range and the main factors affecting the measurement accuracy are analyzed. A novel method of selecting different calibration wavelengths according to COD value to expand the measurement range and the turbidity compensation strategy based on full-spectrum data analysis to improve accuracy in wide measurement range are proposed and realized by an automatic wavelength selection calibration algorithm. Experiments were conducted to verify our idea with the self-developed micro UV–vis spectrophotometer (with a spectral range of 200–750 nm and a resolution of 5 nm) with a 10mm-path-length sample cell. By comparing various algorithms, the Savitzky-Golay (SG) convolution smoothing algorithm and the orthogonal signal correction (OSC) algorithm were chosen to compensate the additional absorption owing to turbidity. The experimental results show that the algorithm can automatically select the optimal characteristic wavelengths according to the spectral data and the measurement range of COD is enormously expanded from 10–150 mg/L to 1–1000 mg/L. The linear correlation coefficient (R²) of model is above 0.9995 and the relative error of measurement is less than 5%. This method can be used for in-situ and online COD measurement.

提出了一种无需试剂的基于紫外吸收光谱的高精度，大范围化学需氧量测量方法。分析了限制测量范围的原因以及影响测量精度的主要因素。提出了一种根据COD值选择不同的校准波长以扩大测量范围的新方法，并提出了一种基于全光谱数据分析的浊度补偿策略，以提高宽测量范围的精度。实验是通过使用自行开发的微型紫外可见分光光度计（光谱范围为200–750 nm，分辨率为5 nm）和路径长度为10mm的样品池来验证的。通过比较各种算法，选择了Savitzky-Golay（SG）卷积平滑算法和正交信号校正（OSC）算法来补偿由于浊度引起的额外吸收。实验结果表明，该算法可以根据光谱数据自动选择最佳特征波长，COD的测量范围从10–150 mg / L极大地扩展到1–1000 mg / L。线性相关系数（R²）模型大于0.9995，测量的相对误差小于5％。此方法可用于原位和在线COD测量。