运用激光诱导击穿光谱(LIBS)技术得到了美国地质勘探局(USGS)系列地质标样的等离子发射光谱。使用人工神经网络测定了不同地质标样的铁元素含量,测得BCR-1G、BHVO-2G、BIR-1G、GSD-1G、GSE-1G标样的铁元素含量与标准含量的相对误差分别为1.86%、5.73%、0.27%、3.86%、2.63%,表明LIBS结合人工神经网络可以很好地测定USGS系列地质标样的铁元素含量。 The plasma emission spectra of a series of geological samples of the US Geological Survey (USGS) were obtained by laser induced breakdown spectroscopy (LIBS). The iron content in different geological samples was measured by artificial neural network. The relative errors between the content of iron and the standard content of BCR-1G, BHVO-2G, BIR-1G, GSD-1G and GSE-1G were 1.86%, 5.73%, 0.27%, 3.86% and 2.63% respectively. It is indicated that the LIBS combined with artificial neural network can well determine the iron content in the USGS geological series samples.