论文部分内容阅读
:提出了反向传播人工神经网络 (BP ANN)参数学习的模拟退火 回火算法。在保持模拟退火算法能收敛于全局最优的特点同时 ,利用模拟回火过程 ,加快了算法的收敛速度。建立了原子吸收光谱单波长下多元测定的BP网络模型。利用干扰效应 ,用常规火焰原子吸收分光光度法在单波长下同时测定了污水中的痕量镉和砷 ,方法简便、安全 ,具有较高的准确度和精密度
: A simulated annealing tempering algorithm for back propagation artificial neural network (BP ANN) parameter learning is proposed. While keeping the simulated annealing algorithm converging to the global optimum, the simulated annealing process is used to speed up the convergence of the algorithm. A BP network model for multivariate determination of atomic absorption spectra at single wavelength was established. Using the interference effect, simultaneous determination of trace amounts of cadmium and arsenic in wastewater by conventional flame atomic absorption spectrophotometry at a single wavelength is simple and safe, with high accuracy and precision