海水硝酸盐原位传感器小型化紫外检测波长优化与仿真研究

面向海水检测行业对小型化、低功耗海水硝酸盐原位传感器的迫切需求，解决传统紫外光谱宽谱扫描方案难以兼顾微型化与检测精度的核心问题。本文构建集成硝酸盐吸收、溴离子等干扰效应及仪器噪声的高保真仿真模型，提出有限最优波长检测策略，结合随机森林算法实现波长智能筛选与非线性建模。根据仿真结果筛选出229nm、230nm、290nm三个波长组合，所建模型在 0.5~1 000 μmol/L 浓度范围内决定系数（R2）达到0.965，残差相对误差约为2.87%；确定仪器信噪比14dB为核心设计门槛。该策略显著降低传感器光源与分光系统复杂度，为基于紫外发光二极管的小型化低功耗传感器奠定了坚实基础，尤其适用于寡营养海域低浓度硝酸盐检测。

英文摘要:

To address the urgent demand for miniaturized and low-power in situ seawater nitrate sensors in the marine monitoring industry, this study tackles the core challenge that traditional broadband UV spectral-scanning approaches struggle to simultaneously achieve compactness and high detection accuracy. We develop a high-fidelity simulation model that integrates nitrate absorption, bromide interference effects, and instrument noise, and propose a “finite optimal wavelength” detection strategy. By combining this strategy with a random forest algorithm, we enable intelligent wavelength selection and nonlinear modeling. Based on simulation results, the wavelength combination of 229 nm, 230 nm, and 290 nm was selected. The resulting model achieved an R2 of 0.965 across a concentration range of 0.5-1 000 μmol/L, with a residual relative error of approximately 2.87%. An instrument signal-to-noise ratio (SNR) of 14 dB was identified as the critical design threshold. This strategy significantly reduces the complexity of the light source and spectral components, providing a solid foundation for compact, low-power sensors based on UV LEDs. It is particularly well suited for detecting low nitrate concentrations in oligotrophic marine environments.

查看全文查看/发表评论下载PDF阅读器