中文摘要：

真空断路器真空度在线检测经历了半个多世纪的研究,现在仍没有很好的技术投入到实际应用。文中采用激光诱导击穿光谱技术（LIBS）,在实验室条件下初步研究了10-3 Pa至105 Pa气压下真空灭弧室屏蔽罩铜材料及周围气体元素元素的LIBS信号,得到了不同真空度下谱线强度随真空度的变化规律。进而通过主成分分析（PCA）和人工神经网络（ANN）算法对多种元素的谱线强度进行分析,获得了主成分因子在相空间的分布和神经网络预测结果。分析结果表明10-3 Pa至105 Pa这10个气压等级下的LIBS信号在PCA相空间中呈现良好的分类聚集,神经网络预测结果准确率达到96.67%。笔者提出了一种真空度在线检测的新方法,克服了传统方法抗电磁干扰能力差、检测限低的难题,具有良好的应用前景。

英文摘要：

Vacuum degree online detection of vacuum interrupter has been investigated for more than 50 years. But, there is still no preferable means can be used to practical application. The spectral information of copper material and ambient gas of vacuum interrupter have been studied from 10-3 Pa to 105 Pa based on Laser-induced breakdown spectroscopy （LIBS） in laboratory and the change rules of spectral intensities against pressure have been obtained. Furthermore, spectral intensities have been analyzed using principal component analysis（PCA） and artificial neural network（ANN）. The PCA result indicates that LIBS signals under ten different pressures from 10-3 Pa to 105 Pa show a good classification. What＇s more, the recognition accuracy can reach up to 96.67% using ANN method. A new online detection method of vacuum degree is proposed in this paper which can overcome the poor electromagnetic interference ability, low limit of detection and hard to realize online detection of tradition methods. This method is of great significant to vacuum degree online detection of vacuum interrupter.