| 基于iPLS-KPCA的高温燃气红外光谱特征提取方法研究 |
| 投稿时间：2024-03-28 修订日期：2024-04-15 点此下载全文 |
| 引用本文： |
| 摘要点击次数: 85 |
| 全文下载次数: 0 |
|
|
| 中文摘要:高温燃气红外光谱特征是判断燃气成分和浓度的有效途径,针对高温燃气红外辐射特性复杂,建模难度高问题,研究一种基于间隔偏最小二乘(iPLS)和核主成分分析(KPCA)相结合的特征提取算法。首先通过iPLS进行预筛选,确定具有最优预测能力的特征光谱波段,避免单个子区间建模过程中有用吸收峰信息的遗失；其次利用核主成分分析降低数据维度,保留贡献率高的关键特征,降低成分预测模型的复杂度。仿真结果表明,经过iPLS-KPCA方法特征提取后,预测模型的复杂度大幅度下降,且预测能力显著提升。 |
| 中文关键词:高温燃气 间隔偏最小二乘 核主成分分析 特征提取 |
| |
| Research on Infrared Spectral Feature Extraction Method for High Temperature Gas Based on iPLS-KPCA |
|
|
| Abstract:The infrared spectral characteristics of high-temperature gas are an effective way to determine the composition and concentration of gas. To address the complex infrared radiation characteristics and high modeling difficulty of high-temperature gas, a feature extraction algorithm based on the combination of the interval Partial Least Squares (iPLS) and the Kernel Principal Component analysis (KPCA) is studied. Firstly, iPLS is used for pre screening to determine the characteristic spectral bands with optimal predictive ability, avoiding the loss of useful absorption peak information during the single sub interval modeling process; Secondly, using kernel principal component analysis to reduce data dimensions, retain key features with high contribution rates, and reduce the complexity of component prediction models. The simulation results show that after feature extraction using the iPLS-KPCA method, the complexity of the prediction model is significantly reduced, and the prediction ability is significantly improved. |
| keywords:High-temperature combustion gas Interval PLS Kernel PCA Feature extraction |
| HTML 查看/发表评论 下载PDF阅读器 |
|
|
|
|
|