论文部分内容阅读
气体绝缘组合电器(GIS)内的绝缘介质SF6及其衍生物的种类和体积严重地影响GIS的绝缘能力,定量分析故障GIS内SF6衍生物有助于评估设备发生故障的原因。为了从GIS设备内SF6气体的红外光谱中获取衍生物的种类和体积,本文将粒子群优化算法和支持向量机回归相结合用于绝缘介质SF6及其部分衍生物的定量分析研究。以衍生物中对绝缘设备具有强腐蚀的气体成分HF、SO2与SF6为研究对象。选取以3种气体红外光谱特征峰为中心,左右各35cm-1波数区域所包含的峰面积为特征值。将获得的13个特征值作为支持向量机的输入,气体浓度作为支持向量机回归的输出。采用粒子群优化算法对支持向量机的参数进行了优化选择,再对优化后的支持向量机进行训练,建立气体分析模型。实验结果表明,采用粒子群算法对支持向量机参数进行优化选择避免了交叉验证法的耗时与盲目性,具有一定的实践意义和应用潜力。
The type and volume of SF6 and its derivatives as insulation medium in gas insulated switchgear (GIS) seriously affect the insulation ability of GIS. Quantitative analysis of SF6 derivatives in fault GIS can help to evaluate the cause of equipment failure. In order to obtain the type and volume of derivatives from infrared spectroscopy of SF6 gas in GIS equipment, particle swarm optimization algorithm and support vector machine regression are combined to study the quantitative analysis of SF6 and its derivatives. The gas components HF, SO2 and SF6 which have strong corrosion on the insulation equipment in the derivatives are the research objects. Select the three kinds of gas infrared spectral peak as the center, left and right 35cm-1 wave number area contains the peak area as the eigenvalue. Thirteen eigenvalues obtained are used as input of support vector machine, and gas concentration is used as output of support vector machine regression. Particle swarm optimization algorithm is used to optimize the parameters of SVM, and then the optimized SVM is trained to establish the gas analysis model. The experimental results show that the particle swarm optimization algorithm can optimize the parameters of SVM to avoid the time-consuming and blindness of cross-validation, which has some practical and application potentials.